Molecular Phylogenetics and Evolution 146 (2020) 106750

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

The evolutionary history of the cellophane bee genus Colletes Latreille T (Hymenoptera: Colletidae): Molecular phylogeny, biogeography and implications for a global infrageneric classification ⁎ Rafael R. Ferraria, , Thomas M. Onuferkoa,b, Spencer K. Moncktona, Laurence Packera a Department of Biology, Faculty of Science, York University, 4700 Keele St., Toronto, ON M3J 1P3, Canada b The Beaty Centre for Species Discovery, Canadian Museum of Nature, Ottawa, ON K1P 6P4, Canada

ARTICLE INFO ABSTRACT

Keywords: Colletes Latreille (Hymenoptera: Colletidae) is a diverse genus with 518 valid species distributed in all biogeo- Apoidea graphic realms, except Australasia and Antarctica. Here we provide a comprehensive dated phylogeny for Colletinae Colletes based on Bayesian and maximum likelihood-based analyses of DNA sequence data of six loci: 28S rDNA, Dated phylogeny cytochrome c oxidase subunit 1, elongation factor-1α copy F2, long-wavelength rhodopsin, RNA polymerase II Geodispersal and wingless. In total, our multilocus matrix consists of 4824 aligned base pairs for 143 species, including 112 Historical biogeography Colletes species plus 31 outgroups (one stenotritid and a diverse array of colletids representing all subfamilies). Systematics Overall, analyses of each of the six single-locus datasets resulted in poorly resolved consensus trees with con- flicting phylogenetic signal. However, our analyses of the multilocus matrix provided strong support forthe monophyly of Colletes and show that it can be subdivided into five major clades. The implications of our phy- logenetic results for future attempts at infrageneric classification for the Colletes of the world are discussed. We propose species groups for the Neotropical species of Colletes, the only major biogeographic realm for which no species groups have been proposed to date. Our dating analysis indicated that Colletes diverged from its sister taxon, Hemicotelles Toro and Cabezas, in the early Oligocene and that its extant lineages began diversifying only in the late Oligocene. According to our biogeographic reconstruction, Colletes originated in the Neotropics (most likely within South America) and then spread to the Nearctic very early in its evolutionary history. Geodispersal to the Old World occurred soon after colonization of the Northern Hemisphere. Lastly, the historical biogeo- graphy of Colletes is analyzed in light of available geological and palaeoenvironmental data.

1. Introduction Kuhlmann et al., 2009; Bystriakova et al., 2018; Ascher and Pickering, 2019). A recent analysis of the environmental factors determining the With 518 valid described species, the cellophane bee genus Colletes global distribution of Colletinae demonstrated that their highest species Latreille (Hymenoptera: Colletidae) is the seventh-most diverse of all richness is in the semi-arid regions located at middle latitudes bee genera and second among colletids, after Hylaeus Fabricius (Ascher (Bystriakova et al., 2018). Contrary to Colletes, the other genera of and Pickering, 2019). The actual species richness of Colletes, however, Colletinae—Hemicotelles Toro and Cabezas (2 spp.), Mourecotelles Toro has been estimated to be closer to 700 (Kuhlmann et al., 2009). The and Cabezas (12 spp.) and Xanthocotelles Toro and Cabezas (11 genus has received a significant amount of taxonomic attention over the spp.)—are all confined to South America, most of them to thesub- past decade resulting in a large number of new species (> 70 spp.) tropical and temperate portions of Chile and Argentina (Moure et al., being described (Kuhlmann, 2014a,b; Kuhlmann and Proshchalykin, 2007, 2012). It is important to emphasize that the generic classification 2011, 2013, 2014a, 2015, 2016; Kuhlmann and Pauly, 2013; of Colletinae is controversial, with the number of genera accepted Proshchalykin and Kuhlmann, 2015; Niu et al., 2013a, b, 2014; Ferrari varying from two to six among authors (Toro and Cabezas, 1977, 1978; and Silveira, 2015; Hall et al., 2016; Balboa et al., 2017; Ferrari, 2017, Michener, 1989, 2007; Moure and Urban, 2002; Moure et al., 2007, 2019). Unlike most colletid genera, which have a comparatively re- 2012; Ascher and Pickering, 2019). stricted geographic distribution, Colletes is found in all biogeographic Colletes are typically robust, mid-sized (8–15 mm) bees with rela- realms except Australasia and Antarctica (Michener, 1989, 2007; tively dense pubescence compared to most other colletids (Michener,

⁎ Corresponding author. E-mail address: [email protected] (R.R. Ferrari). https://doi.org/10.1016/j.ympev.2020.106750 Received 20 September 2019; Received in revised form 27 January 2020; Accepted 28 January 2020 Available online 03 February 2020 1055-7903/ © 2020 Elsevier Inc. All rights reserved. R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

1989, 2007). Morphological synapomorphies supporting the mono- this with the following objectives: (i) to determine the little known phyly of the genus include the form of the subhorizontal surface of the relationships among the NW lineages of Colletes and thus provide a metapostnotum which has a series of longitudinal carinae or sinuous basis for future systematic studies; (ii) to assess the validity of the striae in both sexes (unique within Colletinae), forewings with a sig- currently available subgenera from the OW; and (iii) to reconstruct a moidal second recurrent vein in both sexes (unique among all bees), biogeographic scenario that explains the current distribution of the and T1 subequal in length to T2 in males (unique within Colletinae). genus around the globe in light of geological and palaeoenvironmental Colletes is also monophyletic according to maximum parsimony data. (Kuhlmann et al., 2009) and Bayesian (Almeida and Danforth, 2009; Almeida et al., 2012, 2019) analyses of DNA sequence data; the former 2. Material and methods publication provided the most comprehensive phylogeny for Colletes so far (91 spp.; Kuhlmann et al., 2009) based on two genes (28S rDNA and 2.1. Taxon sampling and specimens studied cytochrome c oxidase subunit 1). With the exception of the subgenus Ptilopoda Friese, which was The subfamilial classification of Colletidae adopted herein follows proposed for an unusual species (C. maculipennis Friese [= C. spilopterus Almeida et al. (2019), in which Paracolletes Smith is a member of Di- Cockerell]) with spotted wings from Central America (Friese, 1921), no phaglossinae, Callomelitta Smith is in its own subfamily, and the other infrageneric classification for Neotropical Colletes has been suggested genera traditionally classified in Paracolletinae (e.g. Michener, 2007, as (Michener, 2007; Kuhlmann et al., 2009). All subgroups proposed Paracolletini) are considered to belong to Neopasiphaeinae. Our taxon within the genus so far—the species groups defined by Noskiewicz, sampling comprises 143 species, 31 of which were sampled as out- Stephen and Kuhlmann as well as all other subgenera (Albocolletes groups and 112 are Colletes (Table S1). Among the outgroups, 22 spe- Warncke, Denticolletes Noskiewicz, Elecolletes Warncke, Nanocolletes cies were chosen to represent the diversity within all colletid sub- Warncke, Pachycolletes Bischoff, Rhinocolletes Cockerell and Simcolletes families based on previous phylogenetic analyses (Magnacca and Warncke)—cover exclusively the Nearctic and/or Old World (OW) Danforth, 2006; Packer, 2008; Almeida and Danforth, 2009; Almeida faunas (Noskiewicz, 1936; Stephen, 1954; Stoeckhert, 1954; Warncke, et al., 2012, 2019). They belong to the following subfamilies: Callo- 1978; Kuhlmann, 2014). All subgenera were subsequently ignored with melittinae (1 sp.), Diphaglossinae (3 spp.), Euryglossinae (3 spp.), Hy- the justification that they had been described for either a single unusual laeinae (4 spp.), Neopasiphaeinae (5 spp.), Scrapterinae (2 spp.) and species (Denticolletes, Ptilopoda and Rhinocolletes) or for artificial groups Xeromelissinae (4 spp.). We also sampled eight species of the other of rather ordinary species (Pachycolletes and Warncke’s subgenera) with genera of Colletinae – Hemicotelles (1 sp.), Mourecotelles (2 spp.) and a restricted geographic distribution in the western Palaearctic Xanthocotelles (5 spp.) – the closest relatives of Colletes (Michener, 1989, (Michener, 1989, 2007). More recently, Kuhlmann et al. (2009) re- 2007; Kuhlmann et al., 2009). The generic classification of Colletinae assessed the status of the available subgenera from the OW and pro- follows Toro and Cabezas (1977, 1978), in which Hemicotelles and posed an updated infrageneric classification for the lineages found Xanthocotelles are recognized as genera (as opposed to subgenera of there, based on their phylogenetic results. Meanwhile, the phylogenetic Mourecotelles; as in Michener, 2007) and Rhynchocolletes Moure is a relationships among the New World (NW) lineages of Colletes remain junior synonym of Colletes (rather than a senior synonym of Mour- poorly understood. The Neotropical Colletes in particular have histori- ecotelles; as in Moure et al., 2007, 2012). Finally, one stenotritid (Ste- cally received much less taxonomic attention compared to those from notritus sp.) was sampled to root the trees given that Stenotridae is the the other biogeographic realms (Michener, 2007: 169). However, a sister taxon of Colletidae (McGinley, 1980; Danforth et al., 2006a, series of recently published revisions have significantly augmented our 2006b; Almeida and Danforth, 2009; Almeida et al., 2012, 2019; knowledge of Neotropical Colletes, thus reducing the taxonomic im- Branstetter et al., 2017; Sann et al., 2018). Among the sampled Colletes, pediment pertaining to them (Toro, 1999; Genaro, 2003; Ferrari and 74 species are from the NW resulting in the most comprehensive sam- Silveira, 2015; Balboa et al., 2017; Ferrari, 2017, 2019). pling of NW fauna of the genus assembled by a phylogenetic study to Resolving the relationships among Neotropical species of Colletes, in date. Of these, 30 species are from the Nearctic, representing 13 of the particular those found in South America where the genus most likely 18 species groups proposed by Stephen (1954) for North American arose (Michener, 1979, 1989, 2007; Kuhlmann et al., 2009; Almeida Colletes (the single representative of the C. impunctatus Nylander group et al., 2012), would be pivotal in attempting to understand the histor- actually belongs in the C. clypearis Morawitz group from the Pa- ical biogeography of the genus and the subfamily as a whole. The hy- laearctic; see Kuhlmann et al., 2009); the other 44 species are from the pothesis that the ancestral Colletes may have inhabited South America Neotropics. With regard to the OW Colletes, we included 38 species, was first suggested by Michener (1979), but only recently has it become ensuring that all major clades recovered in Kuhlmann et al.’s (2009) supported by phylogenetic evidence (Kuhlmann et al., 2009; Almeida phylogeny were represented in our analyses. All subgenera, including et al., 2012). It has been hypothesized that from South America Colletes Rhinocolletes (which was not sampled by those authors), as well as 30 of spread to North America and then to the OW (Kuhlmann et al., 2009). the 40 species groups that have been proposed for the OW Colletes The crown age of Colletes has been dated to the early Oligocene (c. 30 (Noskiewicz, 1936; Kuhlmann et al., 2009; Kuhlmann, 2014b) were million years ago [Mya]; Almeida et al., 2012), which coincides ap- included. proximately with the timing of the complete fragmentation of Gond- The locality, taxonomic and repository information of all studied wana (Upchurch, 2008). Molecular biogeographic studies have re- specimens is listed in Table S2. Most were collected by the authors of vealed that the processes that culminated in the formation of the the present study; however, some were either borrowed from museums current southern continents (Africa, Antarctica, Oceania and South or donated by collaborators (listed under Acknowledgements). The America) played an important role in the diversification of many an- Neotropical Colletinae were identified by the senior author (RF) with imal groups (e.g. ratites [Cooper et al., 2001], chironomids [Krosch available keys (Cockerell, 1913, 1914; Toro and Cabezas, 1977, 1978; et al., 2011], osteoglossiform fishes [Lavoué, 2016], frogs [Feng et al., Genaro, 2003; Ferrari and Silveira, 2015; Balboa et al., 2017) and/or by 2017]), including bees (e.g. megachilids [Litman et al., 2011] and, more comparison with primary types, as part of the revisions by Ferrari specifically, colletids [Almeida et al., 2012, 2019]). (2017, 2019). Identifications of Nearctic Colletes were made by TO and In this study, the evolutionary and biogeographic histories of corroborated by RF with reference to the original descriptions and Colletes are investigated. Here we provide a comprehensive phylogeny available keys (Stephen, 1954; Mitchell, 1960; Ascher and Pickering, for the genus based on Bayesian and maximum likelihood-based (ML) 2019) and examination of the male seventh sternum, which is usually analyses of nuclear and mitochondrial DNA data for 112 species, re- diagnostic. All OW Colletes had been identified by an expert on the presenting groupings from throughout the range of the genus. We do genus (M. Kuhlmann, Kiel University), except for a few species which

2 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750 were identified by RF with the keys for the Eastern European amplifications are in Tables S3 and S4, respectively. (Proshchalykin and Kuhlmann, 2012; Kuhlmann and Proshchalykin, 2014b) and Ethiopian (Kuhlmann and Pauly, 2013) faunas. Species- 2.3. DNA extraction, amplification and sequencing level identification of most specimens was subsequently confirmed through DNA barcoding. The bee specimens from which we extracted DNA were collected in the field, where they were killed and preserved in individual scintilla- 2.2. Gene locus selection tion vials containing absolute ethanol. These were then kept in a −20 °C freezer until the specimens were further processed. In most The dataset used in our phylogenetic analyses consisted of frag- cases, muscle tissue was obtained by grinding the entire specimen’s ments of four nuclear protein-coding genes (elongation factor-1α copy mesosoma with a polypropylene pellet pestle after immersing both in F2 [EF1a], long-wavelength rhodopsin [opsin], RNA polymerase II [pol liquid nitrogen within an Eppendorf tube. To permit subsequent mor- II], and wingless) as well as a 658 bp segment of one mitochondrial phological analysis of rare specimens, we detached the head, pros- protein-coding gene (cytochrome c oxidase subunit 1 [COI]—the DNA ternum, propleura and forelegs and then removed as much muscle barcode region), and one nuclear ribosomal RNA locus (28S rDNA large tissue as possible from the inside of the exposed mesosomal cavity using subunit [28S]). A combination of these loci has been extensively em- extrafine forceps that had previously been flame-sterilized. Theob- ployed in molecular phylogenies of bees over the past two decades tained tissue was then ground as above. Total DNA was extracted using (Leys et al., 2002; Brady and Danforth, 2004; Danforth et al., 2004, a Mag-Bind® Blood DNA HDQ 96 Kit in Professor Amro Zayed’s la- 2006a, 2006b, 2008; Brady et al., 2006, 2011; Larkin et al., 2006; boratory at York University, as outlined in detail in Onuferko et al. Cameron et al., 2007; Almeida et al., 2008, 2012, 2019; Kawakita et al., (2019). 2008; Praz et al., 2008; Almeida and Danforth, 2009; Cardinal et al., The fragments of the five nuclear loci were amplified with thefol- 2010; Flores-Prado et al., 2010; Rasmussen and Cameron, 2010; Litman lowing primers: 28S – Bel28S-For (Belshaw and Quicke, 1997) and et al., 2011, 2016; Payne, 2014; Praz and Packer, 2014; Martins and 28SD4-Rev (Danforth et al., 2006a); EF1a – HaF2For1 and F2-rev1 Melo, 2016; Trunz et al., 2016; Onuferko et al., 2019), including Col- (Danforth et al., 1999); opsin – Opsin-For and Opsin-Rev (Mardulyn and letes (Kuhlmann et al., 2009). Whitfield, 1999); pol II – polfor2a and polrev2a (Danforth et al., Given that the vast majority of the currently available DNA data for 2006a); wingless – Wg-Collet-For (Almeida and Danforth, 2009) and Colletes had been obtained from OW species (see Kuhlmann et al., Lep-Wg2a-Rev (Brower and DeSalle, 1998). The primer sequences are 2009), we focused our efforts on sequencing as many NW species ofthe given in Table S4 and the amplification protocols followed Onuferko genus as possible. In total, we sequenced 63 NW Colletes species for at et al. (2019). We assessed the quality of the crude PCR products by least five of our six targeted genes (Table S1). Additional Colletinae for running 5 μL of each sample in agarose-gel electrophoresis. Good- which a full (or nearly full) dataset was available included eight OW quality, contaminant-free PCR products, represented as single well-de- species of Colletes and eight species from among the other three genera fined bands of the appropriate size, were sent to Bio Basic Inc.for of the subfamily (Table S1). The xeromelissine Chilicola longiceps purification and Sanger sequencing. PCR products were sequenced in (Ashmead), which belongs in the same subgenus (Hylaeosoma Ash- both directions with the same primers used in the PCRs. mead) as the only two known colletid fossils—Ch. gracilis Michener and Poinar and Ch. electrodominica Engel (Michener and Poinar, 1996; 2.4. Sequence assembly and alignment Engel, 1999; Miklasevskaja, 2017)—was also sequenced. Most barcodes used in our analyses are published here for the first time (see Table S3); The resulting two strands of each sequenced sample were assembled the remainder have either been published in a recent taxonomic revi- into single sequences in Sequencher v.5 using the default parameters of sion of the Colletes species found in Eastern South America (Ferrari, the automatic assembling option (Dirty data, with ReAligner, 3′ gap 2019) or as part of a comprehensive phylogenetic study of the asso- placement: minimum overlap 20, minimum match 85%). Final se- ciated cuckoo-bee genus Epeolus Latreille (Hymenoptera: Apidae; quences were then exported to Mega v.5.2.2 (Tamura et al., 2011) Onuferko et al., 2019). Data for all other outgroups and OW species of where they were aligned with the algorithm Muscle v.3.3.2 (Edgar, Colletes were obtained from GenBank. While sequences of all loci were 2004) using default parameters (gap open −400, gap extend 0, lambda available for nearly all outgroups, only COI and 28S sequences were 24). This procedure was done for each gene separately. We trimmed the available for most OW Colletes. Sequences of three South American longest sequence(s) within each single-locus block to eliminate regions species of Colletes – C. bicolor Smith, C. furfuraceus Holmberg and C. sp. that were poorly represented (i.e. present in fewer than half of the 1 – were also obtained from GenBank (see Table S1). sampled taxa). In particular, the sequences of the outgroup species To augment our taxon sampling for Neotropical Colletes, we added obtained from GenBank were often longer than the newly generated DNA barcodes for 15 species from six different countries (Tables S1 and sequences presented herein. Regions of 28S where the alignment was S2), all of which had been assigned separate Barcode Index Numbers, ambiguous were also discarded. The boundaries between exons and which represent operational taxonomic units (OTUs) that closely cor- intron(s) of EF1a (one intron) and opsin (two introns) were assessed by respond with real species (Ratnasingham and Hebert, 2013). The use of comparing our sequences with those of Apis mellifera Linnaeus DNA barcodes to maximize the representability of highly diverse genera (Danforth et al., 2006a, b). Introns of opsin sequences of outgroups in molecular phylogenies has recently been shown to yield useful re- (non-Colletes Colletinae excepted) could not be satisfactorily aligned sults (e.g. Trunz et al., 2016; Onuferko et al., 2019). Due to a lack of and were also discarded. Lastly, the nucleotide sequences were trans- revisions and keys, we could not confirm the species-level identities of lated into peptide sequences to establish correct reading frames and these 15 OTUs and instead identified them by numbers. All previously verify the absence of stop codons in exons. The final length of each obtained DNA barcodes available through collaborators (see Acknowl- single-locus sequence matrix was as follows: 28S (1266 bp), COI edgements) and new DNA barcode sequences were generated at the (657 bp), EF1a (1004 bp), opsin (667 bp), pol II (811 bp) and wingless Canadian Centre for DNA Barcoding (CCDB) at the University of Guelph (419 bp) for a total of 4824 bp. Centre for Biodiversity Genomics (Guelph, Canada) using standard protocols (as in Gibbs, 2009). Resulting DNA barcodes were uploaded 2.5. Data partitioning and undated phylogeny estimation to the Barcode of Life Data System (BOLD) online database (Ratnasingham and Herbert, 2007), where their trace files were in- First, each single-locus matrix was individually exported to spected for quality. The list of the species for which DNA barcodes were Mesquite v.3.6 (Maddison and Maddison, 2018) to be converted into obtained from BOLD and the primer sequences used in the appropriate CFG files for subsequent analyses. The best partitioning

3 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750 scheme and model of DNA evolution was selected for each locus using electrodominica from Miocene amber (Michener and Poinar, 1996; PartitionFinder v.2.1.1 based on the Bayesian Information Criterion Engel, 1999). BEAST 2 was run for 108 generations, with sampling (BIC) using the ‘greedy’ algorithm (Lanfear et al., 2016). The partitions occurring every 103 generations. Resulting trace files were imported and respective models determined for each single-locus matrix were as into Tracer to check for convergence and stationarity and identify the follows: 28S – not partitioned (SYM + Γ + I); COI – 1st and 2nd po- appropriate burn-in. After discarding the trees sampled during the sitions (GTR + Γ + I), 3rd positions (HKY + Γ + I); EF1a – 1st and 2nd burn-in phase (10%), a maximum clade credibility (MCC) tree was positions of exons, 3rd positions of exons, intron (HKY + Γ + I); opsin sampled in TreeAnnotator v.2.5.2 using common ancestor heights – 1st and 2nd positions of exons, 3rd positions of exons, introns (Drummond et al., 2012). The resulting MCC tree was then visualized (K80 + Γ + I); pol II – 1st and 2nd positions, 3rd positions (HKY + G); and edited in FigTree. wingless – 1st and 2nd positions, 3rd positions (GTR + Γ + I). Single-locus trees were then estimated through Bayesian phyloge- 2.7. Biogeographic analysis netic analyses with the models indicated above. All analyses were conducted remotely in the CIPRES Science Gateway v.3.3 (Miller et al., We investigated the historical biogeography of Colletes by re- 2010) using MrBayes v.3.2.6 (Huelsenbeck and Ronquist, 2001; constructing ancestral geographic ranges for its nodes. This approach Ronquist et al., 2012). The number of species included in each analysis allowed us to infer the events (vicariance, geodispersal) that best ex- varied from 73 (pol II) to 130 (COI) depending on the amount of plain the current distribution of Colletes globally. To accomplish this, missing data. We performed two independent MCMC runs (each con- we performed likelihood-based dispersal-extinction-cladogenesis (DEC; taining four chains) for 5 × 107 generations, sampling trees and model Ree et al., 2005; Ree and Smith, 2008) analysis in RASP v.4.1 (Yu et al., parameters every 103 generations. The samples obtained during the 2015) using the package BioGeoBEARS (Matzke, 2013). We did not initial 5 × 106 generations (10%) were discarded as burn-in. Con- incorporate into our DEC analysis the J-parameter, which allows for vergence of the two runs, stationarity of the model parameters and speciation through long-distance geodispersals (Matzke, 2014), for two length of burn-in were assessed in Tracer v.1.7.1 (Rambaut et al., reasons. First, such geodispersal events appear very improbable for 2018a, 2018b). Additional convergence diagnostics included the stan- ground-nesting Colletes which usually fly for short distances (Peakall dard deviation of split frequencies (threshold 10-2), minimal estimated and Schiestl, 2004). Second, Ree and Sanmartín (2018) demonstrated sample sizes (threshold 104) and potential scale reduction factor (equal empirically that the DEC + J model may be highly unparsimonious in or very close to 1). Trees were visualized and annotated in FigTree some biogeographic reconstructions. Ancestral ranges were plotted v.1.4.4 (Rambaut, 2016). Support for the clades indicated in the trees onto the MCC tree from BEAST 2. Herein, five biogeographic areas were was represented by their posterior probabilities, which, in turn, were recognized – (A) Neotropics, (B) Nearctic, (C) Palaearctic, (D) Afro- calculated as the percentage of sampled trees from the stationary dis- tropics and (E) Indo-Malayan – following the boundaries of Olson et al. tribution in which a given clade was recovered. (2001: see Fig. 1). The sources of the geographic data for the Colletinae We also investigated the evolutionary history of Colletes by per- were threefold: (i) comprehensive online databases of bee records forming multilocus phylogenetic analyses. The individual single-locus (Moure et al., 2012; Ascher and Pickering, 2019), (ii) recently pub- matrices were concatenated into a single matrix 4824 bp in length lished revisionary studies of South American Colletes (Ferrari and while maintaining the 13 partitions (and corresponding best models of Silveira, 2015; Ferrari, 2017, 2019) and (iii) discussions with an expert DNA evolution) that had previously been indicated by PartitionFinder. on OW Colletes (M. Kuhlmann, pers. comm.). We constrained the ana- A multilocus Bayesian tree was then estimated in MrBayes following the lyses to allow no ancestral species to occupy more than two areas (as in aforementioned phylogenetic methods. Additionally, we conducted a Voelker, 1999; Praz and Packer, 2014; Trunz et al., 2016). Doing so ML analysis of the multilocus matrix using RAxML v.8 (Stamatakis, ensured that unrealistic combined ranges (e.g. Nearctic plus Indo- 2014) in CIPRES. The ML analysis was performed through 103 boot- Malay) were excluded. Probabilities were set to 1.0 for geodispersal strap (BS) replicates in which the GTR model with a gamma distribution among adjacent areas (e.g. Palaearctic and Indo-Malay). was applied for each partition. 3. Results 2.6. Dated phylogeny estimation 3.1. Power of resolution and phylogenetic signal of different loci A dated multilocus phylogeny for the sampled taxa was estimated in a Bayesian framework, using an uncorrelated lognormal relaxed clock Our results showed that the power of resolution of each of the six model (Drummond et al., 2006) in BEAST 2 (Bouckaert et al., 2014). single-locus analyses (Figs. S1–S6) was relatively low, even though This analysis was also conducted remotely in CIPRES. First, the multi- support for the retrieved clades was moderate to strong overall—the locus matrix was imported into BEAUti v.2.5.2 (Bouckaert et al., 2014) proportion of highly-supported clades (PP 90–100) varied between 53 where the ‘BEAST Model Test’ option was selected for each partition to and 68% among loci. While opsin yielded the most resolved tree (Fig. search for the best models of DNA evolution. Unlike other methods in S4), which contained roughly 81% of the maximum possible number of which the best models are determined a priori, with this option the best internal nodes (68 of 84), the analysis of 28S resulted in a largely un- models are determined simultaneously with phylogenetic estimation. resolved tree (Fig. S1) with merely 49% of the maximum possible Substitution models were unlinked across partitions, whereas the trees number of internal nodes being recovered (59 of 120; note that the and clock models were linked (see Drummond et al., 2015). The ‘Ca- denominator depended on the number of taxa for which sequences were librated Yule Model’ with a random starting tree was defined as the tree available). The analyses of the other four loci yielded intermediate re- prior, and two calibration points were used to time-calibrate the tree, as sults (see Figs. S2, S3, S5 and S6). In general, the nuclear protein-coding follows. The node representing the most recent common ancestor genes and 28S were very efficient at resolving the deeper relationships (MRCA) of Colletidae and Stenotritidae was given a lognormal dis- (including those among the outgroups), whereas COI resolved mostly tribution (log[mean] 4.249, log[SD] 0.153 [which corresponds to a the shallower ones. 95% quantile of 90 Mya, with a median value of 70 Mya]) based on the Overall, the phylogenetic signal was incongruent across single-locus phylogenomic analysis of Apoidea by Sann et al. (2018). A second datasets (Fig. 1). COI (Fig. S1) and 28S (Fig. S2) yielded the most lognormal distribution (log[mean] 3.41, log[SD] 0.247 [which corre- contradictory topologies in comparison to the one resulting from the sponds to a 95% quantile of 45 Mya, with a median value of 30 Mya]) multilocus analysis (Fig. 1), although it should be noted that for c. 35% was assigned for the MRCA of Ch. longiceps and Ch. rostrata (Friese), of the sampled species these were the only data available (Table S1). In based on the estimated age of the fossil species Ch. gracilis and Ch. turn, the phylogenetic signal of both EF1a and pol II were largely

4 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Fig. 1. Bayesian tree obtained through analysis of the multilocus data matrix in MrBayes. Phylogenetic signal of, and topological congruence between, each individual locus with respect to the multilocus analysis are depicted ac- cording to the colour codes shown in the box on the left. ‘Not applicable’ means that topological congruence could not be assessed because taxon sampling for the corresponding locus was insufficient. Numbers below in- ternal branches are posterior prob- abilities of the clades retrieved in the multilocus analysis. The positions of C. acutus, C. cunicularius, C. graeffei and C. sierrensis (grey) are questionable. Boxed numbers from 1 to 5 indicate the five major clades of Colletes. Exemplars il- lustrating some of the diversity of Colletes are: female C. cyaniventris (clade 1), female C. clypeonitens (clade 2), female C. zuluensis (clade 3), female C. halophilus (clade 4) and female C. petropolitanus (clade 5).

congruent with that of the multilocus dataset (Fig. 1; see also Figs. S3 the multilocus dataset (Fig. 1), its monophyly was contradicted by the and S5, respectively). However, an important exception to the afore- analysis of EF1a (Fig. S3). In the latter, Hemicotelles ruizii (Herbst) was mentioned patterns is that, while Colletes was recovered as mono- recovered as the sister species of C. cyanescens (Haliday) plus C. mus- phyletic with maximum support in both analyses of 28S (Fig. S1) and culus Friese, suggesting paraphyly of Colletes (Fig. S3). Additionally, the

5 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Fig. 2. Maximum clade credibility tree inferred through Bayesian analysis of the multilocus data matrix in BEAST 2. Colour-coded, vertical bars on the right depict the infrageneric classification of Colletes proposed in this paper. Numbers below internal branches are posterior probabilities of the corresponding clades. The positions of C. acutus, C. cunicularius, C. graeffei and C. sierrensis (grey) are questionable. Boxed numbers from 1 to 5 at nodes show the five major clades of Colletes; boxes with numbers followed by letters indicate other clades of relevance mentioned in the text. The exemplar at the node of Colletes is a female C. zuluensis.

6 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750 analysis of COI placed the species of Mourecotelles within a large remain unplaced in subgenera, a number of them belong in Pachy- polytomy containing the bulk of the species of Colletes, thus also ren- colletes. Colletes albomaculatus (Lucas) is an isolated taxon in the phy- dering the latter paraphyletic (Fig. S2). However, while analysis of pol logeny—the type species of the subgenus Albocolletes. Colletes alboma- II (Fig. S5) was inconclusive, Colletes was recovered as a strongly-sup- culatus was recovered as the sister species of all Colletes excluding those ported monophyletic group in both separate analyses of opsin (Fig. S4) of clade 1 with maximum support. The second isolated branch corre- and wingless (Fig. S6). sponds to C. acutus, whose relationship to other Colletes remains unclear as the node placing it as sister to clade 4 plus clade 5 was weakly 3.2. Multilocus analyses supported.

The Bayesian (Figs. 1 and 2) and ML (Fig. S7) analyses conducted 3.3. Divergence time estimates and historical biogeography with the multilocus dataset resulted in trees with very similar topolo- gies. The monophyly of Colletes and its placement as sister to Hemi- Our analysis in BEAST 2 (Figs. 2 and S8) suggested that the MRCA of cotelles were strongly supported (PP 100 and BS > 90) in all analyses. Colletinae arose in the early Eocene (stem age 51.2 Mya, 95% highest Together, the two genera were placed as sister to the clade containing probability density [HPD] 68.4–38.2 Mya). The extant colletines, Mourecotelles and Xanthocotelles, which, in turn, were also recovered as however, likely began diversifying only in the middle Eocene (crown reciprocally monophyletic with strong support (PP 100 and BS 92). age 39.9 Mya, 95% HPD 51.2–27.8 Mya). The analysis also suggested There, however, were some topological differences between the Baye- that Colletes and Hemicotelles diverged from one another in the late sian and ML trees (in the following the first and second patterns were Eocene (stem age 35.8 Mya, 95% HPD 47.4–25.7 Mya), and that the those from the Bayesian and ML analyses, respectively): (i) C. fulvipes diversification of extant Colletes began in the early Oligocene (crown Spinola was recovered as sister to either C. chusmiza Rojas and Toro age 30.7 Mya, 95% HPD 40.1–21.4 Mya). The crown ages of all five (Fig. 2) or C. flaminii Moure (Fig. S7); (ii) C. nigropilosus Ferrari formed major clades of Colletes (clades 1–5) date to the early to middle Miocene a monophyletic group with either C. atacamensis Janvier plus C. strigi- (16.7–15.1 Mya), except the South American clade 1, the crown age of nasis Vachal (Fig. 2) or with C. alocochila Moure (Fig. S7); (iii) C. acutus which was placed in the late Oligocene (26.8 Mya, 95% HPD Pérez was placed in either a separate branch (Figs. 1 and 2) or as sister 35.1–18.1 Mya). to the clade containing the largest representatives of Pachycolletes, the The DEC analysis (Fig. 3a) inferred that the MRCA of Colletes members of Nannocolletes plus the Afrotropical Colletes except C. som- probably had a Pan-American distribution (i.e. its geographic range ereni Cockerell (Fig. S7); (iv) the C. fodiens (Fourcroy) group either was included both the Neotropics and Nearctic) during the early Oligocene rendered paraphyletic by C. nasutus Smith plus C. wolfii Kuhlmann (crown age 30.7 Mya). The probability of this ancestral range re- (Fig. 2) or constituted a monophyletic group (Fig. S7); (v) C. annejohnae construction based on our data was 98.5%. In total, six lineage inter- Kuhlmann plus C. roborovskyi Friese were recovered as more closely changes between these two realms were suggested (Fig. 3b): two geo- associated with the C. conradti Noskiewicz plus C. clypearis groups dispersal events from the Neotropics to the Nearctic (crown ages 30.7 (Fig. 2) than with the C. succinctus (Latreille) plus C. caspicus Morawitz and 5.1 Mya), and four events in the opposite direction (crown ages groups (Fig. S7); and (vi) C. perileucus Cockerell was placed as sister to 11.7, 11.2, 7.1 and 6.5 Mya). The DEC analysis also inferred that Col- either the clade containing the C. willistoni Robertson and C. consors letes first colonized the Palaearctic from the Nearctic likely during the Cresson groups (Fig. 2) or to the one which included the C. intermixtus late Oligocene (crown age 26.5 Mya). There were two additional geo- Swenk and C. cunicularius (Linnaeus) groups (Fig. S7). dispersal events each from the Nearctic into the Palaearctic (crown ages Our preferred phylogenetic tree (Fig. 2) indicated that Colletes can 12.4 and 2.0 Mya) and from the latter to the former (crown ages 11.0 be divided into five major clades and two isolated single-species and 10.8 Mya). From the Palaearctic, Colletes spread into the Afro- lineages, as follows. A strongly-supported clade (clade 1) containing tropics twice, once in the middle Miocene (crown age 12.8 Mya) and exclusively South American species was recovered as the sister group to then in the latest Miocene (crown age 5.8 Mya); no geodispersal was all remaining Colletes. Clade 1 encompassed four subclades (clades inferred as having occurred in the opposite direction (Fig. 3b). The 1A–1D) each of which was retrieved with maximum support; none Indo-Malay was probably the last realm to be colonized by Colletes; belong to any of the previously described subgenera of Colletes.A however, we could not verify when this event first occurred because no second, small clade (clade 2) was recovered with maximum support and ancestral range including this realm was reconstructed: rather two of included four species from the Nearctic/northern Neotropics: two of the sampled Palaearctic species have independently extended their them, C. clypeonitens Swenk and C. compactus Cresson, remain unplaced ranges into the Indo-Malay (Fig. 3b). to subgenus at present (the identity of the other two species could not be determined). A third, weakly-supported clade (clade 3) consisted of 4. Discussion three members of Pachycolletes (clade 3A), all sampled Nanocolletes (clade 3B) and all but one of the included Afrotropical species (clade 4.1. Phylogenetic analyses 3C) that belong in none of the currently available subgenera. While clade 3A was relatively weakly supported, clades 3B and 3C as well as In this paper, we provided the most comprehensive phylogeny of the sister-group relationship between the two were recovered with Colletes to date; comprehensive both in terms of number of species in- maximum support. A fourth, relatively weakly supported clade (clade cluded (112 spp., > 20% of the known total) and the amount of data 4) included species from a range of named subgenera. Clade 4A was employed (six loci, 4824 bp in total). In particular, we performed dense well supported and included all species of Elecolletes that we sampled. It taxon sampling regarding the NW fauna of the genus (74 spp., c. 66% of was placed as sister to a weakly-supported clade (clade 4B) which in- the total) which allowed us to shed light on the largely unknown re- cluded, among others, the type species of four different available sub- lationships among these lineages for the first time. All analyses con- genera: C. graeffei Alfken (Denticolletes), C. nasutus (Rhinocolletes), C. ducted with the multilocus data matrix (Figs. 1, 2 and S7) yielded very succinctus (Colletes s. str.) and C. similis Schenk (Simcolletes). Within similar topologies, in which five major clades were recovered. Overall, clade 4B, the species of the C. similis group formed a monophyletic support for these clades varied from weak (e.g. clade 3, posterior group with C. nasutus and C. wolfi (clade 4Bi), which does not include probability [PP] 0.58–0.61 and BS < 50) to maximum (e.g. clade 2). the other species of Simcolletes (placed in clade 4Bii). The fifth clade Some subclades, most notably those within clades 4 and 5, were com- (clade 5), which was recovered with moderate support, consisted solely posed of relatively long terminal branches connected by short inter- of NW Colletes with the exception of C. cunicularius, a widespread spe- nodes (Fig. S8), which likely explained, at least partially, their weak cies in the Palaearctic. While the bulk of species included in clade 5 support (for a detailed discussion on the topic, see Alfaro et al., 2003).

7 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

(caption on next page)

8 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Fig. 3. A pruned maximum clade credibility chronogram (without outgroups) obtained through dated Bayesian analysis of the multilocus matrix in BEAST 2. (A) Pie- charts depict ancestral geographic ranges of corresponding nodes inferred by the BioGeoBEARS analysis under the DEC model in RASP. The positions of C. acutus, C. cunicularius, C. graeffei and C. sierrensis (grey) are questionable. Boxed numbers from 1 to 5 indicate the five major clades of Colletes. (B) Map illustrating the five biogeographic areas (A–E) considered in the analysis; polymorphic ranges are shown within rectangles. Arrows and corresponding numbers in the diagram below the map represent the lineage exchanges among geographic areas inferred by the DEC model; the bee resting on the diagram is a female C. petropolitanus. On the timescale, Pli. = Pliocene; Qua. = Quaternary; Mya = million years ago.

It appears safe to assume that our phylogenetic analyses were, at (2009). For instance, in both studies (i) the C. daleae Cockerell and C. least to some extent, negatively affected by the large amount of missing compactus groups were recovered as the Nearctic groups of earliest di- data (sometimes also referred to as ‘incomplete taxa’; see Kearney, vergence (clade 2); (ii) the Colletes albomaculatus group was placed on 2002; Wiens, 2003) contained in our multilocus matrix. Of the 112 an isolated branch as sister to all Colletes excluding the basalmost Colletes species we sampled, 61 (c. 54%) were sequenced for the nuclear Neotropical lineages; and (iii) the clade containing all but one Afro- protein-coding genes (EF1a, opsin, pol II, wingless) (see Table S1). The tropical species appeared as sister to Nanocolletes (clade 3). Never- reasons for the missing data for the other 51 species were twofold: first, theless, there are some important topological differences between the only 28S and COI data were available for 30 of the 38 OW species of phylogenies proposed in the two studies. For example, the early-di- Colletes, all of them originally sequenced by Kuhlmann et al. (2009), verging South American lineages of Colletes formed a strongly-sup- except C. nasutus (Schmidt et al., 2015). Perhaps this is why clades 3 ported monophyletic group (clade 1) in our multilocus analysis, and 4, in which nearly all OW Colletes were nested, were recovered with whereas the ones sampled by Kuhlmann et al. (2009) came out as a PP 0.61 and 0.84, respectively (Fig. 2). Another explanation would be series of isolated clades near the MRCA of Colletes. Also, the sampled that both clades 3 and 4 are actually not monophyletic and that the members of the C. fodiens group formed a monophyletic group with C. missing data associated with their constituent taxa contributed to the nasutus (Rhinocolletes) and C. wolfi (Colletes s. str.) in our analysis (clade recovery of a spurious topology (see Lemmon et al., 2009). Second, we 4Bi) rather than with the other Simcolletes as in Kuhlmann et al.’s adopted an approach of using DNA barcodes to augment our taxon (2009) phylogeny. Therefore, in light of our phylogenetic results we sampling, as Trunz et al. (2016) did with regards to Megachilini and argue that a new infrageneric classification for Colletes is warranted and Onuferko et al. (2019) to Epeolus. If on the one hand this approach should be the subject of a future study (Table S6). allowed us to include a series of Neotropical species from otherwise poorly-sampled regions (most notably tropical Mexico, see Tables S1 and S2), on the other it added substantially to the amount of missing 4.2.1. Clade 1 data. Nevertheless, the use of DNA barcodes in combination with Clade 1 is an exclusively South American one which included, as far slowly-evolving nuclear genes resulted in a mostly well-supported as we could determine, 67 of the 124 valid species (c. 54%) of Colletes phylogeny that permitted us both to propose an updated infrageneric found on the continent (for an updated checklist, see Table S5). Clade 1 classification for Colletes (Fig. 2; Section 4.2, below) and carry out a was, in turn, comprised of four subclades (1A–1D), none of which fit in more accurate assessment of its historical biogeography (Fig. 3; Section any of the currently available subgenera (Fig. 2); therefore, four new 4.3, below). subgenera will need to be described for the species in these subclades in a future systematic study (Table S6). To further convey the taxonomic complexity within clade 1, we recommend the use of 15 species groups 4.2. Infrageneric classification (Table S5). As argued by Kuhlmann et al. (2009), species groups are informal, yet useful, alternatives to describing large numbers of sub- Constructing a phylogeny with a comprehensive sampling of the genera which can be redefined whenever further phylogenetic evidence Neotropical Colletes permitted us to, for the first time, allocate all spe- is available. This approach has been adopted in the taxonomy of other cies currently accepted as valid in the realm to species groups (Table species-rich bee genera, such as Euglossa Latreille (Dressler, 1978, S5). All 19 species groups proposed herein include exclusively Neo- 1982a, 1982b, 1982c; Roubik, 2004; Ferrari et al., 2017), Nomada tropical Colletes, although a significant proportion of other species Scopoli (Alexander, 1994) and Epeolus (Onuferko et al., 2019). found in this realm belong to groups that were originally described for It is very likely that the type species of Rhynchocolletes (C. albicinctus the Nearctic (Stephen, 1954; see also Table S5). Placement of the [Moure]) would have been placed within Colletes had the species been Neotropical species into species groups was done based on the results of included in our phylogenetic analysis. Colletes albicinctus possesses two our preferred phylogenetic analysis (Fig. 2) in combination with a of the three morphological features that distinguish Colletes from the comparative study of the male S7, which has been widely recognized as other genera of Colletinae—the basal area of the metapostnotum the most useful feature for species delimitation and identification bearing a series of short longitudinal carinae in both sexes, and T1 within Colletes (e.g. Mitchell, 1960; Kuhlmann, 2007; Kuhlmann and longer than broad dorsally in males. This is why Rhynchocolletes is Proshchalykin, 2011; Proshchalykin and Kuhlmann, 2012; Ferrari, herein seen as a synonym of Colletes (as in Michener, 1989, 2007). In C. 2017). Although the shape of the male S7 tends to be species-specific albicinctus, however, the second recurrent vein of the forewing is only throughout the genus, interspecific variation in the structure is not slightly arcuate posteriorly, rather than strongly sigmoidal as it is in random and relatively similar forms are usually shared by sets of clo- almost all species of the genus (it appears to have also reversed to its sely-related species (Noskiewicz, 1936; Stephen, 1954; Kuhlmann et al., ancestral condition in several other species that are unambiguously 2009). Relevant characteristics of the male S7 used by us to delineate members of the genus, such as C. atripes Smith [RF, pers. obs.]). It the 19 Neotropical groups included, for example, the relative length of moreover seems probable that C. albicinctus is a member, or the sister the neck in relation to the disc, general shape of the disc (e.g. sub- species, of either clade 1A or clade 1C given that in the males the triangular, subquadrate), density and distribution of short pubescence, metasomal terga (most notably T1–T4) are metallic dark blue, a very and absence/presence of laterobasal expansions and medially-directed rare feature within Colletes only exhibited (to the best of our knowl- folds (following mostly Stephen, 1954). For a number of species (mostly edge) by members of those two clades. If this conjecture proves correct from Argentina, Central America and tropical Mexico) only the original through further analysis, the nominal taxon Rhynchocolletes will be descriptions were available and, therefore, their placement is only available to name one of the two South American clades at the sub- tentative (indicated by question marks in Table S5). generic level. The major relationships within Colletes found herein (Fig. 2) were relatively similar to those previously reported by Kuhlmann et al.

9 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

4.2.2. Colletes albomaculatus of our clade 3A (as in Kuhlmann et al., 2009). Colletes albomaculatus is the type species of the subgenus Albocolletes which includes an additional three Palaearctic species: C. alfkeni 4.2.6. Clade 4 Noskiewicz, C. dorsalis Morawitz and C. punctatus Mocsáry (Kuhlmann, Clade 4 included two subclades: all representatives of Elecolletes, 2000), none of which were available for sequencing. In our multilocus which formed a relatively strongly-supported monophyletic group analysis, C. albomaculatus was recovered as not belonging to any of the (clade 4A, PP 0.92) and its sister group (i.e. clade 4B; Fig. 2). Although five major clades that constituted the phylogenetic treeof Colletes; in- the type species of Elecolletes, C. elegans Noskiewicz, could not be in- stead it was placed (with maximum support; Fig. 2) on an isolated cluded in our analysis, we included C. omanus Kuhlmann, which is the terminal branch sister to all of its congeners, except those of clade 1. most probable sister species of C. elegans (Kuhlmann et al., 2009). Thus, the maintenance of the subgeneric status of Albocolletes as cur- Elecolletes is herein circumscribed (Table S6) exactly as it was by rently understood (Table S6) is supported by the present study. Kuhlmann et al. (2009), containing the C. carinatus Radoszkowski, C. caspicus, C. hylaeiformis Eversmann, C. nigricans Gistel and C. squamosus 4.2.3. Clade 2 Morawitz groups (Table S6). All sampled species of Elecolletes are Clade 2 is a small, exclusively North American group which does not confined to the Palaearctic, except C. somereni (C. squamosus group), fit into any of the currently available subgenera of Colletes (Fig. 2). which is Afrotropical (Kuhlmann, 1998; Ascher and Pickering, 2019). Therefore, a new subgenus is needed to accommodate its constituent Classification of clade 4B was particularly challenging as it included species (Table S6). Clade 2 had maximum support and included the the type species of four subgenera. Furthermore, nodal support for its Nearctic C. compactus and C. clypeonitens (the former is the sole member constituent clades was very weak overall (Fig. 2). Within clade 4B, the of the C. compactus group while the second belongs in the C. daleae members of the C. fodiens group (including C. similis, type species of group which contains nine valid species (Stephen, 1954)) as well as two Simcolletes) formed a monophyletic group (clade 4Bi) with the type and unidentified species from tropical Mexico. In Kuhlmann et al.’s (2009) single species of Rhinocolletes (C. nasutus) but without the remaining phylogeny, both C. compactus and C. clypeonitens were recovered as species of Simcolletes (Fig. 2). Kuhlmann et al. (2009) predicted that C. more closely related to Neotropical Colletes, however, they were here nasutus (which was not included in their analysis) would be a close shown to be closer to the OW species belonging to clade 3 (Fig. 2). relative of C. albomaculatus, presumably based on morphological data. However, none of our analyses confirmed this prediction; instead they 4.2.4. Clade 3 showed that C. albomaculatus and C. nasutus belong to distantly-related Clade 3 consisted of three major monophyletic groups (Fig. 2): one lineages within Colletes (Figs. 1, 2 and S7). Here, we suggest that Sim- included three of the largest species of Pachycolletes (clade 3A), another colletes be synonymized with Rhinocolletes; the latter would then include the representatives of Nanocolletes (clade 3B), and the third all Afro- the following Palaearctic species groups: the C. nasutus, C. anchusae tropical species, except C. somereni (clade 3C). There currently are no Noskiewicz (Colletes s. str.) and C. fodiens (Simcolletes) groups, and subgeneric nominal taxa available for clade 3A and clade 3C, thus both possibly the C. senilis (Eversmann), C. tardus Noskiewicz and C. uralensis will need to be described in the future. Our phylogenetic analyses Noskiewicz groups (Simcolletes). However, the sister-group relationship provided further support for the status of Nanocolletes as proposed by between C. daviesanus Smith plus C. nasustus and C. wolfi was very Kuhlmann et al. (2009), that is, containing only the C. diodontus Be- poorly supported (PP 0.17). Therefore, it would also be sensible to noist, C. foveolaris Pérez and C. nanus Friese groups (see Table S6). maintain Simcolletes as a valid subgenus while expanding Rhinocolletes It, however, should be noted that the placement of C. acutus and C. to also include the C. anchusae group. cunicularius outside clade 3A by our Bayesian phylogenetic analyses Within clade 4Bii, C. succinctus and C. hederae Schmidt and Westrich (Figs. 1 and 2) contradicts morphological evidence, which strongly (Colletes s. str.) formed a monophyletic group with the type and single suggests placement of the two species within that clade. Hence, future species of Denticolletes, C. graeffei (Fig. 2). Based on this result, we phylogenetic studies, preferably of an integrative total evidence ap- suggest that the monotypic Denticolletes be synonymized with Colletes s. proach, should aim to resolve this discrepancy (see further discussion str. We must emphasize, however, that the nodal support for this ar- on this topic, below). rangement was very low (PP 0.31). Moreover, C. graeffei shares derived morphological features with members of clade 4Bi, particularly C. na- 4.2.5. Colletes acutus sutus and C. wolfi (see Proshchalykin and Kuhlmann, 2012). Therefore, The second isolated terminal branch in our tree was C. acutus which, the actual position of C. graeffei within the phylogenetic tree of Colletes together with C. acutiformis Noskiewicz, composes the Palaearctic C. requires further investigation. If its sister-group relationship with the acutus group (Noskiewicz, 1936; Kuhlmann, 2000). Colletes acutus has members of the C. succinctus group is confirmed in the future, then historically been classified within Pachycolletes (Warncke, 1978; Denticotelles will have to be synonymized under Rhinocolletes rather Kuhlmann et al., 2009), but our analysis showed that it does not in fact than with Colletes s. str. belong to any of the currently available subgenera of Colletes (Fig. 2). In turn, the clade containing C. succinctus/C. hederae plus C. graeffei Even though it may appear sensible to recommend the description of a was recovered as sister to the one which included C. annejohnae (C. new subgenus to accommodate the members of the C. acutus group, the mixtus Radoszkowski group), C. roborovskyi (C. roborovskyi group) and nodal support for the recovered arrangement (i.e. C. acutus as sister to all species currently classified within Simcolletes, except those of the C. clade 4 plus clade 5; see Fig. 2) was very weak. Thus, we opted to place similis group. If our classificatory scheme is to be followed, then Colletes both C. acutus and C. acutiformis as incertae sedis (Table S6). s. str. must be expanded to also include the C. mixtus and C. roborovskyi It is noteworthy that in a previous comprehensive phylogeny of groups (neither is currently placed in a subgenus), and the C. clypearis, Colletes (Kuhlmann et al., 2009), C. acutus was recovered as sister to the C. conradti, C. marginatus Smith and C. simulans Cresson groups of clade containing the members of the C. formosus Pérez, C. inaequalis Say Simcolletes (Table S6). Lastly, it is important to note that the position of and C. lacunatus Dours groups (which herein constituted clade 3A), the Palaearctic C. sierrensis Frey-Gessner (C. marginatus group) as sister although with low support (BS < 50). This hypothesis is nonetheless to the Nearctic C. phaceliae Cockerell (C. hyalinus Provancher group) is further supported by there being significant morphological similarity questionable in light of morphological data. Colletes sierrensis shares among them, including in the male S7 (M. Kuhlmann, pers. comm.). seemingly derived characteristics associated with the male S7 with Therefore, more data (particularly from slowly-evolving loci) are members of the C. succinctus group, particularly to its nominal species, needed to verify whether C. acutus belongs to a separate branch within and thus should in theory belong to a branch of earlier divergence the phylogenetic tree of Colletes (as inferred by our analyses), or, within clade 4Bii, potentially as sister to all Nearctic species (M. whether it actually comprises a monophyletic group with the members Kuhlmann, pers. comm.).

10 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

4.2.7. Clade 5 et al., 2009). In our phylogenetic analyses, however, the two groups Clade 5 included the bulk of the sampled Pachycolletes (including its were recovered as reciprocally monophyletic (Figs. 1, 2 and S7), which type species, the Palaearctic C. cunicularius) although most of its con- supported a single early geodispersal event into the Nearctic. In theory, stituent species are not currently placed in any of the available sub- two possible geodispersal routes could have been used by Colletes to the genera (Tables S5 and S6). Colletes cunicularius was recovered as more Nearctic: either via the Isthmus of Panama (IP) or the Antilles. The closely related to some of the exclusively NW groups of Colletes (par- complete formation of the IP, and the consequent closure of the Central ticularly to the C. intermixtus group) than to the Palaearctic ones that American Seaway (CAS), most likely took place 4–3 Mya (Coates and are currently classified within Pachycolletes (see Fig. 2 and Table S6). Stallard, 2013). Whereas we estimated the crown age of the MRCA of However, as previously noted, the position of C. cunicularius within our Colletes to be c. 31 Mya (Fig. 3a), a geodispersal through the IP thus phylogenetic trees should be interpreted with caution when mor- appears improbable. However, more recent studies have unraveled a phology is considered: its male S7 appears to share some derived fea- rather complex geological history according to which a version of the IP tures with that of the members of the C. inaequalis group, particularly C. may have formed much earlier than currently thought, either in the thoracicus Smith and C. validus Cresson (see Stephen, 1954; Kuhlmann middle Miocene (Montes et al., 2015) or near the Oligocene-Miocene and Proshchalykin, 2011), thus suggesting that C. cunicularius would boundary (Bacon et al., 2015). Even less conservative estimates have instead belong to clade 3A. suggested the IP was present as a chain of islands during the late Eocene The analysis also showed, for the first time, that various Nearctic (Montes et al., 2012). Thus, the possibility of stepping-stone geodis- groups proposed by Stephen (1954) should be considered to belong to persal through islands where the IP is currently located cannot be ruled Pachycolletes; these are: the C. hyalinus, C. intermixtus, C. latitarsis Ro- out. Alternatively, the MRCA of Colletes could have geodispersed to the bertson, C. nudus Robertson, C. productus Robertson, C. robertsonii Dalla Nearctic via the Antilles which, in spite of their gradual movement Torre and C. willistoni groups (Table S6). Kuhlmann et al. (2009) had eastwards since the late Eocene, have maintained a fairly uniform already proposed that the C. americanus Cresson and C. consors groups configuration over the past 40 Ma(Graham, 2018). This scenario, belong to Pachycolletes. Also, three of the Neotropical species groups however, is weakened by the fact that the MRCA of all Colletes species proposed herein that are not members of Clade 1—C. bombiformis Metz, currently found in the Antilles – C. granpiedrensis Genaro, C. hicaco C. petropolitanus Dalla Torre and C. rugicollis Friese—also belong in Genaro, C. montefragus Raw and C. submarginatus Cresson (Raw, 1984; Pachycolletes according to our phylogenetic results (Table S6). Genaro, 2001, 2003; Ascher and Pickering, 2019) – is of fairly recent Based on the morphology of the male S7, it is likely that the divergence (crown age 3.9 Mya; see clade 5 in Fig. 3a). There are nu- Neotropical C. spilopterus Cockerell (the type species of Ptilopoda) would merous examples of pre-3.5 Mya geodispersals across the CAS by ter- have been placed as sister to, or even within, the C. latitarsis group had restrial animals documented in the literature (compiled by Bacon et al., the species been available and included in our analysis (see Michener, 2015), including bees (Hines, 2008; Ramírez et al., 2010; Rasmussen 1954: Fig. 1 and Stephen, 1954: Fig. 10). If this conjecture is supported and Cameron, 2010). Although it seems reasonable to consider that the in future phylogenetic studies, the members of clade 5 will have to be MRCA of Colletes could have colonized the Nearctic in the early Oli- regarded as belonging to Colletes (Ptilopoda). gocene, it is nonetheless highly unlikely that a continuous Pan-Amer- ican distribution could have been maintained, as suggested by our DEC 4.3. Divergence times and historical biogeography analysis (Fig. 3a), given the tenuous connection between North and South America at that time. Our DEC analysis indicated that the divergence between Colletes and Four late Miocene, back geodispersals into the Neotropics from the Hemicotelles in the late Eocene (stem age 35.8 Mya) took place some- Nearctic were inferred by our biogeographic reconstruction (Fig. 3b). where in the Neotropics (Fig. 3a). Whereas both Hemicotelles and the The earliest of these were associated with the MRCA of C. intermixtus earliest-diverging lineage of Colletes (clade 1) are confined to South plus C. latitarsis (crown age 11.7 Mya), and that of the C. nudus group America (Fig. 3a), it is very likely that the two genera diverged within (crown age 11.2 Mya); whereas the most recent ones were associated that continent. The fact that their closest relatives (i.e. Mourecotelles and with the MRCA of C. bombiformis plus C. gilensis Cockerell (crown age Xanthocotelles) are also South American further supports this hypoth- 7.1 Mya), and with that of the C. compactus group (crown age 6.5 Mya). esis. A South American origin of Colletes has long been suspected This time span coincided with the formation and expansion of sa- (Michener, 1979, 1989, 2007), although it has only recently been vannah-like and grassland biomes in the Nearctic west, which gradually supported by phylogenetic evidence (Kuhlmann et al., 2009; Almeida replaced mixed mesophytic forest that had dominated most of the et al., 2012). Colletes likely arose at the onset of a major Antarctic landscape from 35 to 10 Mya (Pound et al., 2012; Meseguer et al., glaciation event—the Eocene-Oligocene transition—which resulted 2015). This change in vegetation was likely boosted by the drastic from a prolonged period of global cooling initiated in the middle Eo- transition from the warm Middle Miocene Climatic Optimum to the cene (Zachos et al., 2001, 2008; Houben et al., 2012; Mudelsee et al., glaciation event that lasted from 15 to 13 Mya (Zachos et al., 2008). By 2014; Pound and Salzmann, 2017). As a consequence, tropical forest the beginning of the late Miocene, the open-vegetation biomes ex- that covered South America almost entirely in the Palaeocene began to tended from the North American Great Plains to what is now south- be gradually replaced with Nothofagus-dominated forest at high lati- western Mexico (Pound et al., 2012; Meseguer et al., 2015) and possibly tudes, with open-vegetation biomes arising towards the centre of the constituted the geodispersal route used by these four lineages of Colletes continent (Iglesias et al., 2011; Meseguer et al., 2015). The combination to reach the Neotropics. The same route may have been taken by the of these biotic and abiotic factors likely played an important role in the MRCA of C. latitarsis and C. submarginatus (crown age 5.1 Mya) in the diversification of the Colletes lineages that remained isolated in South opposite direction. Intriguingly, no lineage interchange between the America. two realms was inferred for the time period corresponding to the Great Arguably the most unexpected result of our DEC analysis was the American Biotic Interchange (i.e. after 3 Mya; Marshall, 1979, 1982; inference of a Pan-American distribution for the MRCA of Colletes Web, 2006; Woodburne, 2010), although this event lies within the 95% (Fig. 3a). The reconstruction implied a very early geodispersal (crown HDP for the last three geodispersals. age 30.7 Mya) to the Nearctic from the Neotropics during the early The Andes are the longest and second highest cordillera on the Oligocene. Based on their findings, Kuhlmann et al. (2009) suggested planet (Graham, 2009), extending along almost the entire north-south that the Nearctic species groups of earliest divergence – C. dalaea and C. axis of South America, from Venezuela to southern Chile and Argentina compactus groups – would be descendants of two different Neotropical (Morrone, 2018). Not surprisingly, they have contributed to diversifi- ancestors; in other words, that the Nearctic would have been colonized cation of both plant (e.g. Antonelli et al., 2009; Struwe et al., 2009; at least twice independently by Colletes in its early evolution (Kuhlmann Strelin et al., 2017) and animal (e.g. Brumfield and Edwards, 2007;

11 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Weir and Price, 2011; Blandin and Purser, 2013; Almeida et al., 2019) association with the C. inaequalis group (clade 3A) than with the C. taxa through vicariance. The Andes uplift has drastically changed the intermixtus group (clade 5). Thus, both the evolutionary and biogeo- precipitation regime (Rech et al., 2006, 2010), atmospheric CO2 con- graphic histories of C. cunicularius (and its closest allies) require further tent (Graham, 2009) and western fluvial system (Hoorn et al., 1995) in investigation. Later, two Palaearctic ancestors would have reached the South America, ultimately driving continental landscape evolution over Nearctic virtually at the same period in the late Miocene: the MRCA of the last c. 65 Ma (Hoorn et al., 2010; Coudurier-Curveur et al., 2015) the C. inaequalis group (crown age 11.0 Mya), and that of C. impunctatus and providing the conditions for the diversification of an endemic biota plus C. phaceliae (crown age 10.8 Mya). At that time, the only available (Morrone, 2018). Our ancestral-range reconstruction detected three route connecting the two realms would have been the BLB (Sanmartín cladogenetic events within Colletes that could potentially be linked to et al., 2001; Brikiatis, 2014; Graham, 2018). Along with the high lati- the Andean orogeny, namely, at the nodes corresponding to the MRCAs tude regions across the Holarctic, the BLB would have been almost of C. arthuri Ferrari plus C. ferenudus Ferrari (crown age 11.4 Mya), C. entirely covered with boreal forest, which by then had largely replaced rufipes Smith plus C. atacamensis (crown age 9.2 Mya), and C. cyaneus the Oligocene mixed mesophytic forest (Pound et al., 2012; Meseguer Holmberg plus C. cyanescens Spinola (crown age 6.8 Mya). Even though et al., 2015; Graham, 2018). It should be noted that if C. cunicularius is the Andes may have begun rising as early as 100 Mya (Cobbold et al., sister to the C. inaequalis group, as suggested by morphological data, 2007; Armijo et al., 2015), the most pronounced uplift period did not then the most parsimonious biogeographic explanation for clade 3A occur until c. 12–11 Mya (Hoorn et al., 2010; Jordan et al., 2014), would be a single geodispersal of the MRCA of the C. inaequalis group which could potentially explain why the trans-Andean sister groups are from the Palaearctic into the Nearctic. The latest geodispersal event was considerably more recent. inferred for the MRCA of C. phaceliae plus C. sierrensis (crown age Based on our biogeographic reconstruction, Colletes would have first 2.0 Mya), which may have geodispersed from the Nearctic to the Pa- geodispersed to the Palaearctic from the Nearctic in the late Oligocene. laearctic across the BLB in the early Pleistocene. The BLB was likely This event was inferred for the MRCA of all Colletes except clade 1 already exposed in the late Pliocene, and although it became sub- (crown age 26.5 Mya), which suggested geodispersal to the OW soon merged in the earliest Pleistocene, it was repeatedly exposed during after colonization of the Northern Hemisphere (crown age 30.7 Mya). that period as a result of a series of glaciation events (Sanmartín et al., The Bering Land Bridge (BLB) which had been connecting Alaska and 2001; Ehlers and Gibbard, 2007). The highly derived position of the Siberia, at least intermittently, since the late Cretaceous (Sanmartín Palaearctic species C. sierrensis within clade 4Bii, as sister to the et al., 2001) was likely the route used by Colletes. The estimated arrival Nearctic C. phaceliae, is also doubtful if one takes morphology into of the genus in the Palaearctic coincides very closely with a relatively consideration. As previously noted, the morphology of the male S7 of C. warm period (c. 27–25 Mya) that preceded a major glaciation event in sierrensis points to a closer association with the C. succinctus group, the Oligocene-Miocene boundary (Mudelsee et al., 2014). Given its high rather than with the C. hyalinus and C. simulans groups (M. Kuhlmann, latitude and position relative to the Arctic Circle, climatic conditions in pers. comm.), as suggested by our analyses (Figs. 1, 2 and S7). In bio- the BLB during the late Oligocene might have nonetheless been unlikely geographic terms, such an arrangement would imply that C. sierrensis is to permit geodispersal of terrestrial organisms (Sanmartín et al., 2001; instead a descendant of a Palaearctic (not Nearctic) ancestor and that Zachos et al., 2001, 2008), particularly ectotherms such as bees. lineage exchanges between the two realms ended in the mid-late Mio- However, the BLB has been acknowledged to have served as an im- cene boundary (c. 11–10 Mya), not in the early Pleistocene (c. 2 Mya) as portant refugium for boreal species during the various Pleistocene inferred by our biogeographic reconstruction (Fig. 3). glaciations (DeChaine, 2008) and so perhaps it acted similarly during Our biogeographic reconstruction indicated that the Afrotropical earlier cold periods. Two alternative, trans-Atlantic geodispersal routes species of Colletes are descendants of at least two Palaearctic ancestors may also have been available at that time: the Thulean Bridge, which that geodispersed to the Afrotropics in the Miocene (Fig. 3a), as pre- connected the Queen Elizabeth Islands to southern Europe, and the De viously suggested by Kuhlmann et al. (2009). The earliest arrival was Geer Bridge, which united the Canadian Arctic Archipelago and the inferred for the MRCA of C. escalerai Noskiewicz/C. nanus plus C. Fennoscandian peninsula, both through Greenland (Brikiatis, 2014 and antecessus Cockerell/C. aureocinctus Cockerell (crown age 12.6 Mya) references therein). The former had a warmer climate due to its which would have invaded the Afrotropics in the middle Miocene. southern position in relation to the latter (~45°N and > 70°N, re- According to our analysis, this ancestor would have later given rise to spectively) and, consequently, has been widely acknowledged as a more the bulk of the species that inhabit the realm at present (Fig. 3a). Most likely geodispersal route for terrestrial organisms (e.g. Sanmartín et al., of these are endemic to the South African Greater Cape Floristic Region, 2001; Irving, 2005; Praz and Packer, 2014). The Thulean and De Geer which has been detected as the globally most important centre of spe- Bridges are thought to have existed until the early (c. 50 Mya) and late cies richness of Colletes in relative terms (Bystriakova et al., 2018). The (c. 40 Mya) Eocene, respectively (Sanmartín et al., 2001; Condamine most recent arrival in the Afrotropics was found at the node corre- et al., 2013), both too early for Colletes geodispersal. However, it is sponding to the MRCA of C. somereni plus C. merceti Noskiewicz/C. possible that they may have persisted as island chains up until the early omanus (crown age 5.9 Mya), which likely arrived in the realm in the Miocene (Baskin and Baskin, 2016), which could have allowed for latest Miocene. It is possible that both events would have occurred via a stepping-stone geodispersal of cold-adapted Colletes. long, arid corridor encompassing central Asia, the Arabian Peninsula Lineage exchanges between the NW and OW appear to have com- and northern Africa, areas which have been mostly covered with pletely ceased from c. 25–13 Mya, but since then Colletes geodispersed grasslands and savannahs since the Miocene (Senut et al., 2009; between the two at least twice in each direction (Fig. 3a). The earliest Meseguer et al., 2015), a biogeographic scenario that was also discussed among the four was found at the node corresponding to the MRCA of by Kuhlmann et al. (2009). With the increase of aridification initiated the Eurasian species C. cunicularius plus its NW closest relatives, which earlier in the Miocene (Meseguer et al., 2015), these biomes came to would have geodispersed from the Nearctic to the Palaearctic near the replace mixed mesophytic forest (Asia) and tropical forest (Africa), middle-late Miocene boundary (crown age 12.4 Mya). This finding likely facilitating southward geodispersal of Colletes. corroborated the first biogeographic scenario regarding the origin of C. Unfortunately, no ancestral range including the Indo-Malay was cunicularius hypothesized by Kuhlmann et al. (2009), which implied a inferred in our biogeographic analysis, which prevented us from in- single geodispersal event of the ancestor of C. cunicularius from the ferring when Colletes first colonized this realm (Fig. 3a). This was likely Nearctic to the Palaearctic. However, as noted previously, the position a result of us having only two species from this realm, C. esakii Hira- of C. cunicularius in our tree should be interpreted with caution when shima and C. lacunatus, which are primarily Palaearctic in distribution the available morphological evidence is considered (see Stephen, 1954; (Kuhlmann, 1998, 2000; Kuhlmann et al., 2009; Ascher and Pickering, Kuhlmann and Proshchalykin, 2011), which suggests a closer 2019) and appear to have reached the Indo-Malay only more recently.

12 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

This hypothesis is supported by two facts: first, most of their geographic specimens preserved in ethanol. Julio Genaro and Berry Brosi are ranges lie within the Palaearctic, and their known occurrence records thanked for generously allowing us to use the DNA barcodes of the within the Indo-Malay are close to the border between the two realms Cuban and Costa Rican species of Colletes, respectively. RF is particu- (Ascher and Pickering, 2019; M. Kuhlmann, pers. comm.). Second, the larly indebted to Julio Rivera, Eduardo Passuni, Claudio Salas, ancestral ranges of the MRCAs which C. esakii and C. lacunatus share Romualdo Hernández, Alfredo Ugarte, Eduardo Almeida, André with their respective sister species were inferred to include only the Nemésio and Antonio Aguiar for facilitating fieldwork in South Palaearctic (Fig. 3a). It is nonetheless certain that the Indo-Malayan America, and to Javier Cañote, Marina Pucci, Stephanie Téssier, Marcos realm has been invaded at least five times independently by Palaearctic Ramos, Heron Hilário and Felipe Freitas for the help provided in the ancestral Colletes given that representatives of five species groups of field. Some of the specimens included in this study we collected at unequivocal Palaearctic origin – C. caspicus, C. clypearis, C. lacunatus, C. Rondeau Provincial Park, and we thank the Ministry of Natural marginatus and C. succinctus groups – are known to occur there Resources and Forestry for giving us permission to sample bees there. (Kuhlmann, 1998, 2000; Kuhlmann et al., 2009; Ascher and Pickering, We also thank the Texas Parks and Wildlife Department for issuing TO a 2019). permit to collect bees at the Chaparral Wildlife Management Area and area manager Stephen Lange and assistant area manager Sarah 5. Conclusion Resendez for their much-appreciated assistance. The photographs of the Colletes species shown in Figs. 1 and 2 were taken with an imaging In accordance with previous studies, our phylogenetic analyses system purchased through a Canadian Foundation for Innovation and provided further support for the monophyly of Colletes and for its sister- Ontario Foundation for Innovation awards through Canadensys. RF group relationship to Hemicotelles. The approach of using DNA barcodes thanks CAPES (BEX 11875/13-5) and the Faculty of Science of York to augment our taxon sampling allowed us to consolidate the basis for University for the scholarships provided. TO was supported financially future classification attempts regarding the Colletes of the world. Even by a Susan Mann Dissertation Scholarship (granted by the Faculty of though our preferred phylogeny (Fig. 2) was mostly well supported, Graduate Studies at York University), and a Beaty Postdoctoral Fel- more data (especially from slowly-evolving loci) are needed to verify lowship for Species Discovery (granted by the Canadian Museum of whether the poorly-supported groupings recognized herein are truly Nature with funding provided by the Ross Beaty Family). This study monophyletic. More specifically, it will be important to verify whether was funded by the grants awarded to LP by the Natural Sciences and (i) the C. acutus group belongs in none of the currently available sub- Engineering Research Council of Canada (NSERC) and National Geo- genera, (ii) the members of the C. similis group are more closely related graphic. Finally, we thank Michael Kuhlmann and an anonymous re- to C. nasutus rather than to the other groups traditionally considered in viewer for their constructive and extensive feedback on a previous draft Simcolletes, and (iii) C. graeffei is sister to the C. succinctus group. Of of this manuscript. particular relevance is that we placed the large majority of the Neo- tropical Colletes into a monophyletic suite of species groups for the first Appendix A. Supplementary material time. Nevertheless, the position of a number of species from the NW Colletes remains unresolved and therefore further work is needed. Supplementary data to this article can be found online at https:// Our attempts to reconstruct the historical biogeography of Colletes doi.org/10.1016/j.ympev.2020.106750. at a global scale revealed with high confidence that the genus origi- nated somewhere within the Neotropics, most likely within South References America, about 36 Mya. The biogeographic scenario described herein includes an early geodispersal to the Nearctic (c. 31 Mya) followed by Alexander, B.A., 1994. Species-groups and cladistic analysis of the cleptoparasitic bee an early arrival in the Palaearctic (26.5 Mya). From there, Colletes co- genus Nomada. Univ. Kans. Sci. Bull. 55, 175–238. Alfaro, M.E., Zoller, S., Lutzoni, F., 2003. Bayes or bootstrap? A simulation study com- lonized both the Afrotropic and Indo-Malay realms; the former was paring the performance of Bayesian Markov chain Monte Carlo sampling and boot- likely invaded for the first time about 13 Mya but, unfortunately, we strapping in assessing phylogenetic confidence. Mol. Biol. Evol. 20, 255–266. https:// could not date the latter event. Future studies should aim to elucidate doi.org/10.1093/molbev/msg028. Almeida, E.A.B., Danforth, B.N., 2009. Phylogeny of colletid bees (Hymenoptera: which of the South American biogeographic dominions (sensu Morrone, Colletidae) inferred from four nuclear genes. Mol. Phylogenet. Evol. 50, 290–309. 2014) corresponds to the point of origin of Colletes, and when the genus https://doi.org/10.1016/j.ympev.2008.09.028. first colonized the Indo-Malay. Almeida, E.A.B., Packer, L., Danforth, B.N., 2008. Phylogeny of the Xeromelissinae (Hymenoptera: Colletidae) based upon morphology and molecules. Apidologie 39, 75–85. https://doi.org/10.1051/apido:2007063. CRediT authorship contribution statement Almeida, E.A.B., Packer, L., Melo, G.A.R., Danforth, B.N., Cardinal, S.C., Quinteiro, F.B., Pie, M.R., 2019. The diversification of neopasiphaeine bees during the Cenozoic Rafael R. Ferrari: Conceptualization, Methodology, Formal ana- (Hymenoptera: Colletidae). Zool. Scr. 48, 226–242. https://doi.org/10.1111/zsc. 12333. lysis, Investigation, Data curation, Writing - original draft, Almeida, E.A.B., Pie, M.R., Brady, S.G., Danforth, B.N., 2012. Biogeography and di- Visualization. Thomas M. Onuferko: Methodology, Investigation, Data versification of colletid bees (Hymenoptera: Colletidae): emerging patterns fromthe curation, Writing - review & editing. Spencer K. Monckton: southern end of the world. J. Biogeogr. 39, 526–544. https://doi.org/10.1111/j. 1365-2699.2011.02624.x. Investigation, Writing - review & editing. Laurence Packer: Validation, Antonelli, A., Nylander, J.A., Persson, C., Sanmartín, I., 2009. Tracing the impact of the Resources, Writing - review & editing, Supervision, Funding acquisi- Andean uplift on Neotropical plant evolution. Proc. Nat. Acad. Sci. 106, 9749–9754. tion. https://doi.org/10.1073/pnas.0811421106. Armijo, R., Lacassin, R., Coudurier-Curveur, A., Carrizo, D., 2015. Coupled tectonic evolution of Andean orogeny and global climate. Earth-Sci. Rev. 143, 1–35. https:// Acknowledgements doi.org/10.1016/j.earscirev.2015.01.005. Ascher, J.S., Pickering, J., 2019. Discover Life bee species guide and world checklist (Hymenoptera: Apoidea: Anthophila). http://www.discoverlife.org/mp/20q?guide= We are profoundly grateful to Amro Zayed for allowing us to utilize Apoidea_species (accessed 4 February 2019). his lab space, equipment and supplies to extract and amplify most DNA Bacon, C.D., Silvestro, D., Jaramillo, C., Smith, B.T., Chakrabarty, P., Antonelli, A., 2015. samples used in the elaboration of this study. Some past and current Biological evidence supports an early and complex emergence of the Isthmus of members of his team, most notably Alivia Dey and Ida Conflitti, gen- Panama. Proc. Nat. Acad. Sci. 112, 6110–6115. https://doi.org/10.1073/pnas. 1423853112. erously assisted RF, TO and SM during the aforementioned procedures. Balboa, C., Hinojosa-Diaz, I., Ayala, R., 2017. New dark species of the bee genus Colletes Jonathan Huang provided RF and TO with many hours of training in (Hymenoptera, Colletidae) from Mexico and Guatemala. Zootaxa 4320, 401–425. various molecular techniques. We thank Michael Kuhlmann, Petr https://doi.org/10.11646/zootaxa.4320.3.1. Baskin, J.M., Baskin, C.C., 2016. Origins and relationships of the mixed mesophytic forest Bogusch and Claus Rasmussen for the kind donation of Colletes

13 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

of Oregon-Idaho, China, and Kentucky: Review and Synthesis. Ann. Missouri Bot. Dressler, R.L., 1978. An infrageneric classification of Euglossa, with notes on some fea- Gard. 101, 525–552. https://doi.org/10.3417/2014017. tures of special taxonomic importance (Hymenoptera; Apidae). Rev. Biol. Trop. 26, Belshaw, R., Quicke, D.L., 1997. A molecular phylogeny of the Aphidiinae (Hymenoptera: 187–198. Braconidae). Mol. Phylogenet. Evol. 7, 281–293. https://doi.org/10.1006/mpev. Dressler, R.L., 1982a. New species of Euglossa. II. (Hymenoptera: Apidae). Rev. Biol. Trop. 1996.0400. 30, 121–129. Blandin, P., Purser, B., 2013. Evolution and diversification of Neotropical butterflies: Dressler, R.L., 1982b. New species of Euglossa. III. The bursigera species group Insights from the biogeography and phylogeny of the genus Morpho Fabricius, 1807 (Hymenoptera: Apidae). Rev. Biol. Trop. 30, 131–140. (Nymphalidae: Morphinae), with a review of the geodynamics of South America. Dressler, R.L., 1982c. New species of Euglossa. IV. The cordata and purpurea species Trop. Lepid. Res. 23, 62–85. groups (Hymenoptera: Apidae). Rev. Biol. Trop. 30, 141–150. Bouckaert, R., Heled, J., Kühnert, D., Vaughan, T., Wu, C.-H., Xie, D., Suchard, M.A., Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics Rambaut, A., Drummond, A.J., 2014. BEAST 2: A Software Platform for Bayesian and dating with confidence. PLoS Biol. 4, e88. https://doi.org/10.1371/journal.pbio. Evolutionary Analysis. PLoS Comput. Biol. 10, e1003537. https://doi.org/10.1371/ 0040088. journal.pcbi.1003537. Drummond, A.J., Rambaut, A., Bouckaert, R., Xie, W., 2015. Divergence Dating Tutorial Brady, S.G., Danforth, B.N., 2004. Recent intron gain in elongation factor 1-α of colletid with BEAST 2.2.x. http://beast2-dev.github.io/beast-docs/beast2/ bees (Hymenoptera: Colletidae). Mol. Biol. Evol. 21, 691–696. https://doi.org/10. DivergenceDating/DivergenceDatingTutorial.html (accessed 26 February 2019). 1093/molbev/msh062. Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with Brady, S.G., Sipes, S., Pearson, A., Danforth, B.N., 2006. Recent and simultaneous origins BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. https://doi.org/10.1093/ of eusociality in halictid bees. Proc. R. Soc. B 273, 1643–1649. https://doi.org/10. molbev/mss075. 1098/rspb.2006.3496. Edgar, R.C., 2004. MUSCLE: multiple sequence alignment with high accuracy and high Brady, S.G., Litman, J.R., Danforth, B.N., 2011. Rooting phylogenies using gene dupli- throughput. Nucleic Acids Res. 32, 1792–1797. https://doi.org/10.1093/nar/ cations: an empirical example from the bees (Apoidea). Mol. Phylogenet. Evol. 60, gkh340. 295–304. https://doi.org/10.1016/j.ympev.2011.05.002. Ehlers, J., Gibbard, P.L., 2007. The extent and chronology of Cenozoic Global Glaciation. Branstetter, M.G., Danforth, B.N., Pitts, J.P., Faircloth, B.C., Ward, P.S., Buffington, M.L., Quat. Int. 164, 6–20. https://doi.org/10.1016/j.quaint.2006.10.008. Gates, M.W., Kula, R.R., Brady, S.G., 2017. Phylogenomic insights into the evolution Engel, M.S., 1999. A new xeromelissine bee in Tertiary amber of the Dominican Republic of stinging wasps and the origins of ants and bees. Curr. Biol. 27, 1019–1025. https:// (Hymenoptera: Colletidae). Insect Syst. Evol. 30, 453–458. https://doi.org/10.1163/ doi.org/10.1016/j.cub.2017.03.027. 187631200X00561. Brikiatis, L., 2014. The De Geer, Thulean and Beringia routes: key concepts for under- Feng, Y.J., Blackburn, D.C., Liang, D., Hillis, D.M., Wake, D.B., Cannatella, D.C., Zhang, standing early Cenozoic biogeography. J. Biogeogr. 41, 1036–1054. https://doi.org/ P., 2017. Phylogenomics reveals rapid, simultaneous diversification of three major 10.1111/jbi.12310. clades of Gondwanan frogs at the Cretaceous-Paleogene boundary. Proc. Natl. Acad. Brower, A.V.Z., DeSalle, R., 1998. Patterns of mitochondrial versus nuclear DNA sequence Sci. USA 114, 5864–5870. https://doi.org/10.1073/pnas.1704632114. divergence among nymphalid butterflies: the utility of wingless as a source ofchar- Ferrari, R.R., 2017. Taxonomic revision of the species of Colletes Latreille, 1802 acters for phylogenetic inference. Insect Mol. Biol. 7, 73–82. https://doi.org/10. (Hymenoptera: Colletidae: Colletinae) found in Chile. Zootaxa 4364, 1–137. https:// 1046/j.1365-2583.1998.71052.x. doi.org/10.11646/zootaxa.4364.1.1. Brumfield, R.T., Edwards, S.V., 2007. Evolution into and out of the Andes: aBayesian Ferrari, R.R., 2019. A revision of Colletes Latreille (Hymenoptera: Colletidae: Colletinae) analysis of historical diversification in Thamnophilus antshrikes. Evolution 61, from Brazil, Paraguay and Uruguay. Zootaxa 4606, 1–91. https://doi.org/10.11646/ 346–367. https://doi.org/10.1111/j.1558-5646.2007.00039.x. zootaxa.4606.1.1. Bystriakova, N., Griswold, T., Ascher, J.S., Kuhlmann, M., 2018. Key environmental de- Ferrari, R.R., Nemésio, A., Silveira, F.A., 2017. Phylogeny of the orchid-bee genus terminants of global and regional richness and endemism patterns for a wild bee Euglossa Latreille (Hymenoptera: Apidae), with emphasis on the subgenera E. subfamily. Biodivers. Conserv. 27, 287–309. https://doi.org/10.1007/s10531-017- (Glossura) Cockerell and E. (Glossuropoda) Moure. Zool. Anz. 267, 82–100. https:// 1432-7. doi.org/10.1016/j.jcz.2017.02.005. Cameron, S.A., Hines, H.M., Williams, P.H., 2007. A comprehensive phylogeny of the Ferrari, R.R., Silveira, F.A., 2015. The species of Colletinae (Hymenoptera: Colletidae) in bumble bees (Bombus). Biol. J. Linn. Soc. 91, 161–188. https://doi.org/10.1111/j. the Brazilian state of Minas Gerais. Zootaxa 3914, 245–274. https://doi.org/10. 1095-8312.2007.00784.x. 11646/zootaxa.3914.3.2. Cardinal, S., Straka, J., Danforth, B.N., 2010. Comprehensive phylogeny of apid bees Flores-Prado, L., Flores, S.V., McAllister, B.F., 2010. Phylogenetic relationships among reveals the evolutionary origins and antiquity of cleptoparasitism. Proc. Natl. Acad. tribes in Xylocopinae (Apidae) and implications on nest structure evolution. Mol. Sci. USA 107, 16207–16211. https://doi.org/10.1073/pnas.1006299107. Phylogenet. Evol. 57, 237–244. https://doi.org/10.1016/j.ympev.2010.06.019. Coates, A.G., Stallard, R.F., 2013. How old is the Isthmus of Panama? Bull. Mar. Sci. 89, Friese, H., 1921. Nachtrag zur Bienenfauna von Costa Rica. Stett. Entomol. Ztg. 82, 801–813. https://doi.org/10.5343/bms.2012.1076. 73–98. Cobbold, P.R., Rossello, E.A., Roperch, P., Arriagada, C., Gómez, L.A., Lima, C., 2007. Genaro, J.A., 2001. Especies nuevas de abejas de Cuba y la Española (Hymenoptera: Distribution, timing, and causes of Andean deformation across South America. Geol. Colletidae, Megachilidae, Apidae). Rev. Biol. Trop. 49, 951–959. Soc. Lond. Spec. Publ. 272, 321–343. https://doi.org/10.1144/GSL.SP.2007.272. Genaro, J.A., 2003. El género Colletes en Cuba (Hymenoptera: Apiformes: Colletidae). 01.17. Solenodon 3, 53–56. Cockerell, T.D.A., 1913. Descriptions and records of bees.—XLIX. Ann. Mag. Nat. Hist. 11, Gibbs, J., 2009. New species in the Lasioglossum petrellum species group (Hymenoptera: 185–195. https://doi.org/10.1080/00222931308693306. Halictidae) identified with an integrative taxonomic approach. Can. Entomol. 141, Cockerell, T.D.A., 1914. Bees from Ecuador and Peru. J. N. Y. Entomol. Soc. 22, 306–328. 371–396. https://doi.org/10.4039/n09-020. https://www.jstor.org/stable/25003639. Graham, A., 2009. The Andes: a geological overview from a biological perspective. Ann. Condamine, F.L., Sperling, F.A., Kergoat, G.J., 2013. Global biogeographical pattern of Missouri Bot. Gard. 96, 371–386. https://doi.org/10.3417/2007146. swallowtail diversification demonstrates alternative colonization routes inthe Graham, A., 2018. Land Bridges: Ancient Environments, Plant Migrations, and New World Northern and Southern hemispheres. J. Biogeogr. 40, 9–23. https://doi.org/10.1111/ Connections. University of Chicago Press, Chicago. j.1365-2699.2012.02787.x. Hall, H.G., Almquist, D.T., Ascher, J.S., 2016. A new species of Colletes (Hymenoptera: Cooper, A., Lalueza-Fox, C., Anderson, S., Rambaut, A., Austin, J., Ward, R., 2001. Apoidea: Colletidae) from northern Florida and Georgia, with notes on the Colletes of Complete mitochondrial genome sequences of two extinct moas clarify ratite evolu- those states. J. Melittology 58, 1–20. https://doi.org/10.17161/jom.v0i58.4993. tion. Nature 409, 704–707. https://doi.org/10.1038/35055536. Hines, H.M., 2008. Historical biogeography, divergence times, and diversification pat- Coudurier-Curveur, A., Lacassin, R., Armijo, R., 2015. Andean growth and monsoon terns of bumble bees (Hymenoptera: Apidae: Bombus). Syst. Biol. 57, 58–75. https:// winds drive landscape evolution at SW margin of South America. Earth Planet. Sc. doi.org/10.1080/10635150801898912. Lett. 414, 87–99. https://doi.org/10.1016/j.epsl.2014.12.047. Hoorn, C., Guerrero, J., Sarmiento, G.A., Lorente, M.A., 1995. Andean tectonics as a cause Danforth, B.N., Brady, S.G., Sipes, S.D., Pearson, A., 2004. Single-copy nuclear genes for changing drainage patterns in Miocene northern South America. Geology 23, recover Cretaceous-age divergences in bees. Syst. Biol. 53, 309–326. https://doi.org/ 237–240. https://doi.org/10.1130/0091-7613(1995) 023<0237:ATAACF>2.3. 10.1080/10635150490423737. CO;2. Danforth, B.N., Eardley, C., Packer, L., Walker, K., Pauly, A., Randrianambinintsoa, F.J., Hoorn, C., Wesselingh, F.P., ter Steege, H., Bermudez, M.A., Mora, A., Sevink, J., 2008. Phylogeny of Halictidae with an emphasis on endemic African Halictinae. Sanmartin, I., Sanchez-Meseguer, A., Anderson, C.L., Figueiredo, J.P., Jaramillo, C., Apidologie 39, 86–101. https://doi.org/10.1051/apido:2008002. Riff, D., Negri, F.R., Hooghiemstra, H., Lundberg, J., Stadler, T., Sarkinen, T., Danforth, B.N., Fang, J., Sipes, S., 2006a. Analysis of family-level relationships in bees Antonelli, A., 2010. Amazonia through time: Andean uplift, climate change, land- (Hymenoptera: Apiformes) using 28S and two previously unexplored nuclear genes: scape evolution, and biodiversity. Science 330, 927–931. https://doi.org/10.1126/ CAD and RNA polymerase II. Mol. Phylogenet. Evol. 39, 358–372. https://doi.org/ science.1194585. 10.1016/j.ympev.2005.09.022. Houben, A.J., van Mourik, C.A., Montanari, A., Coccioni, R., Brinkhuis, H., 2012. The Danforth, B.N., Sauquet, H., Packer, L., 1999. Phylogeny of the bee genus Halictus Eocene-Oligocene transition: Changes in sea level, temperature or both? Palaeogeogr. (Hymenoptera: Halictidae) based on parsimony and likelihood analyses of nuclear Palaeoclimatol. Palaeoecol. 335, 75–83. https://doi.org/10.1016/j.palaeo.2011.04. EF-1α sequence data. Mol. Phylogenet. Evol. 13, 605–618. https://doi.org/10.1006/ 008. mpev.1999.0670. Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogeny. Danforth, B.N., Sipes, S., Fang, J., Brady, S.G., 2006b. The history of early bee diversi- Bioinformatics 17, 754–755. https://doi.org/10.1093/bioinformatics/17.8.754. fication based on five genes plus morphology. Proc. Natl. Acad. Sci. USA103, Iglesias, A., Artabe, A.E., Morel, E.M., 2011. The evolution of Patagonian climate and 15118–15123. https://doi.org/10.1073/pnas.0604033103. vegetation from the Mesozoic to the present. Biol. J. Linn. Soc. 103, 409–422. DeChaine, E.G., 2008. A bridge or a barrier? Beringia’s influence on the distribution and https://doi.org/10.1111/j.1095-8312.2011.01657.x. diversity of tundra plants. Plant Ecol. Divers. 1, 197–207. https://doi.org/10.1080/ Irving, E., 2005. The role of latitude in mobilism debates. Proc. Nat. Acad. Sc. 102, 17550870802328660. 1821–1828. https://doi.org/10.1073/pnas.0408162101.

14 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Jordan, T.E., Kirk-Lawlor, N.E., Blanco, N., Rech, J.A., Cosentino, N.J., 2014. Landscape the Great American Interchange. Science 204, 272–279. https://doi.org/10.1126/ modification in response to repeated onset of hyperarid paleoclimate states since14 science.204.4390.272. Ma, Atacama Desert, Chile. Geol. Soc. Am. Bull. 126, 1016–1046. https://doi.org/10. Marshall, L.G., Webb, S.D., Sepkoski, J.J., Raup, D.M., 1982. Mammalian evolution and 1130/B30978.1. the great American interchange. Science 215, 1351–1357. https://doi.org/10.1126/ Kawakita, A., Ascher, J.S., Sota, T., Kato, M., Roubik, D.W., 2008. Phylogenetic analysis science.215.4538.1351. of the corbiculate bee tribes based on 12 nuclear protein-coding genes (Hymenoptera: Martins, A.C., Melo, G.A., 2016. The New World oil-collecting bees Centris and Epicharis Apoidea: Apidae). Apidologie 39, 163–175. https://doi.org/10.1051/apido:2007046. (Hymenoptera, Apidae): molecular phylogeny and biogeographic history. Zool. Scr. Kearney, M., 2002. Fragmentary taxa, missing data, and ambiguity: mistaken assumptions 45, 22–33. https://doi.org/10.1111/zsc.12133. and conclusions. Syst. Biol. 51, 369–381. https://doi.org/10.1080/ Matzke, N.J., 2013. Probabilistic historical biogeography: new models for founder-event 10635150252899824. speciation, imperfect detection, and fossils allow improved accuracy and model- Krosch, M.N., Baker, A.M., Mather, P.B., Cranston, P.S., 2011. Systematics and biogeo- testing. Front. Biogeogr. 5, 242–248. https://doi.org/10.21425/F5FBG19694. graphy of the Gondwanan Orthocladiinae (Diptera: Chironomidae). Mol. Phylogenet. Matzke, N.J., 2014. Model selection in historical biogeography reveals that founder-event Evol. 59, 458–468. https://doi.org/10.1016/j.ympev.2011.03.003. speciation is a crucial process in island clades. Syst. Biol. 63, 951–970. https://doi. Kuhlmann, M., 1998. Lectotype designation and new synonymy for Afrotropical and org/10.1093/sysbio/syu056. Oriental bees of the genus Colletes Latr. (Hymenoptera, Apidae, Colletinae). Linz. McGinley, R.J., 1980. Glossal morphology of the Colletidae and recognition of the Biol. Beitr. 30, 559–577. Stenotritidae at the family level (Hymenoptera: Apoidea). J. Kans. Entomol. Soc. 53, Kuhlmann, M., 2000. Katalog der paläarktischen Arten der Bienengattung Colletes Latr., 539–552. https://www.jstor.org/stable/25084069. mit Lectotypenfestlegungen, neuer Synonymie und der Beschreibung von zwei neuen Meseguer, A.S., Lobo, J.M., Ree, R., Beerling, D.J., Sanmartín, I., 2015. Integrating fossils, Arten (Hymenoptera: Apidae: Colletinae). Linz. Biol. Beitr. 32, 155–193. phylogenies, and niche models into biogeography to reveal ancient evolutionary Kuhlmann, M., 2007. Revision of the bees of the Colletes fasciatus-group in southern Africa history: The case of Hypericum (Hypericaceae). Syst. Biol. 64, 215–232. https://doi. (Hymenoptera: Colletidae). Afr. Invertebr. 48, 121–165. org/10.1093/sysbio/syu088. Kuhlmann, M., 2014a. Colletes kinabalu n. sp., first record of the genus for the Malay Michener, C.D., 1954. Bees of Panamá. Bull. Am. Mus. Nat. Hist. 104, 1–176. http://hdl. Archipelago and Southeast Asia (Hymenoptera: Anthophila: Colletidae). J. Melittol. handle.net/2246/1183. 28, 1–6. https://doi.org/10.17161/jom.v0i28.4614. Michener, C.D., 1979. Biogeography of the bees. Ann. Missouri Botanical Garden 66, Kuhlmann, M., 2014b. Order Hymenoptera, family Colletidae. Supplementary records of 277–347. https://doi.org/10.2307/2398833. the genus Colletes Latreille. In: van Harten, A. (Ed.), Arthropd fauna of the UAE, Michener, C.D., 1989. Classification of American Colletinae (Hymenoptera, Apoidea). Volume5. Department of the President's Affairs, Abu Dhabi, pp. 649–656. Univ. Kans. Sci. Bull. 53, 22–703. Kuhlmann, M., Almeida, E.A.B., Laurenne, N., Quicke, D.L.J., 2009. Molecular phylogeny Michener, C.D., 2007. The Bees of the World, second ed. The Johns Hopkins University and biogeography of the bee genus Colletes Latreille, 1802 (Hymenoptera: Apiformis: Press, Baltimore. Colletidae), based on mitochondrial COI and nuclear 28S nuclear sequence data. Michener, C.D., Poinar Jr, G., 1996. The known bee fauna of the Dominican amber. J. Insect Syst. Evol. 40, 291–318. https://doi.org/10.1163/ Kans. Entomol. Soc. 69, 353–361. https://www.jstor.org/stable/25085730. 139956009X12475840653733. Miklasevskaja, M., 2017. Phylogeny of Chilicola (Hylaeosoma) (Hymenoptera: Colletidae) Kuhlmann, M., Pauly, A., 2013. The bee genus Colletes Latreille 1802 in Ethiopia Based on Morphological and Morphometric Data of Extant and Fossil Species, with (Hymenoptera: Apoidea: Colletidae). Zootaxa 3693, 267–292. https://doi.org/10. Description of Seven New Species. M.Sc. thesis. University, Toronto, York. 11646/zootaxa.3693.2.9. Miller, M.A., Pfeiffer, W., Schwartz, T., 2010. Creating the CIPRES Science Gatewayfor Kuhlmann, M., Proshchalykin, M.Y., 2011. Bees of the genus Colletes Latreille 1802 of the inference of large phylogenetic trees. In: Proceedings of the GatewayComputing Asian part of Russia, with keys to species (Hymenoptera: Apoidea: Colletidae). Environments Workshop. IEEE Computer Society, New Orleans, pp. 1–8. Zootaxa 3068, 1–48. https://doi.org/10.11646/zootaxa.3068.1.1. Mitchell, T.B., 1960. Bees of the Eastern United States: Volume I. NC. Agric. Exp. Stn. Kuhlmann, M., Proshchalykin, M.Y., 2013. The genus Colletes (Hymenoptera: Apoidea: Tech. Bull. 141, 1–538. Colletidae) in Central Asia. Zootaxa 3750, 401–449. https://doi.org/10.11646/ Montes, C., Cardona, A., Jaramillo, C., Pardo, A., Silva, J.C., Valencia, V., Ayala, C., Pérez- zootaxa.3750.5.1. Angel, L.C., Rodriguez-Parra, L.A., Ramirez, V., Niño, H., 2015. Middle Miocene Kuhlmann, M., Proshchalykin, M.Y., 2014a. New and little known bees of the genus closure of the Central American Seaway. Science 348, 226–229. https://doi.org/10. Colletes Latreille 1802 (Hymenoptera: Colletidae) from Central Asia. Zootaxa 3846, 1126/science.aaa2815. 591–596. https://doi.org/10.11646/zootaxa.3846.4.8. Montes, C., Cardona, A., McFadden, R., Morón, S.E., Silva, C.A., Restrepo-Moreno, S., Kuhlmann, M., Proshchalykin, M.Y., 2014b. The bees of the genus Colletes Latreille 1802 Ramírez, D.A., Hoyos, N., Wilson, J., Farris, D., Bayona, G.A., Jaramillo, C.A., of the European part of Russia, with keys to species (Hymenoptera: Apoidea: Valencia, V., Bryan, J., Flores, J.A., 2012. Evidence for middle Eocene and younger Colletidae). Zootaxa 3878, 201–247. https://doi.org/10.11646/zootaxa.3878.3.1. land emergence in central Panama: Implications for Isthmus closure. Bull. Geol. Soc. Kuhlmann, M., Proshchalykin, M.Y., 2015. New and remarkable Asian and North African Am. 124, 780–799. https://doi.org/10.1130/B30528.1. species of Colletes Latreille (Hymenoptera: Colletidae). Zootaxa 4028, 81–101. Morrone, J.J., 2014. Biogeographical regionalisation of the Neotropical region. Zootaxa https://doi.org/10.11646/zootaxa.4028.1.3. 3782, 1–110. https://doi.org/10.11646/zootaxa.3782.1.1. Kuhlmann, M., Proshchalykin, M.Y., 2016. The bees of the genus Colletes Latreille Morrone, J.J., 2018. Evolutionary Biogeography of the Andean Region. CRC Press, Boca (Hymenoptera: Colletidae) of the Caucasus region. Zootaxa 4161, 367–385. https:// Raton. doi.org/10.11646/zootaxa.4161.3.5. Moure, J.S., Urban, D., 2002. Catálogo de Apoidea da região Neotropical (Hymenoptera, Lanfear, R., Frandsen, P.B., Wright, A.M., Senfeld, T., Calcott, B., 2016. PartitionFinder 2: Colletidae). III. Colletini. Rev. Bras. Zool. 19, 1–30. https://doi.org/10.1590/S0101- New methods for selecting partitioned models of evolution for molecular and mor- 81752002000100001. phological phylogenetic analyses. Mol. Biol. Evol. 34, 772–773. https://doi.org/10. Moure, J.S., Urban, D., Melo, G.A.R., 2007. Colletini Lepeletier, 1841. In: Moure, J.S., 1093/molbev/msw260. Urban, D., Melo, G.A.R. (Eds.), Catalogue of Bees (Hymenoptera, Apoidea) in the Larkin, L.L., Neff, J.L., Simpson, B.B., 2006. Phylogeny ofthe Callandrena subgenus of Neotropical Region. Sociedade Brasileira de Entomologia, Curitiba, pp. 677–691. Andrena (Hymenoptera: Andrenidae) based on mitochondrial and nuclear DNA data: Moure, J.S., Urban, D., Melo, G.A.R., 2012. Colletini Lepeletier, 1841. In: Moure, J.S., polyphyly and convergent evolution. Mol. Phylogenet. Evol. 38, 330–343. https:// Urban, D., Melo, G.A.R. (Org.), Catalogue of Bees (Hymenoptera, Apoidea) in the doi.org/10.1016/j.ympev.2005.10.003. Neotropical Region - online version. http://www.moure.cria.org.br/catalogue (ac- Lavoué, S., 2016. Was Gondwanan breakup the cause of the intercontinental distribution cessed 15 January 2019). of Osteoglossiformes? A time-calibrated phylogenetic test combining molecular, Mudelsee, M., Bickert, T., Lear, C.H., Lohmann, G., 2014. Cenozoic climate changes: A morphological, and paleontological evidence. Mol. Phylogenet. Evol. 99, 34–43. review based on time series analysis of marine benthic δ18O records. Rev. Geophys. https://doi.org/10.1016/j.ympev.2016.03.008. 52, 333–374. https://doi.org/10.1002/2013RG000440. Lemmon, A.R., Brown, J.M., Stanger-Hall, K., Lemmon, E.M., 2009. The effect of am- Niu, Z.Q., Kuhlmann, M., Zhu, C.D., 2013a. A review of the Colletes succinctus-group biguous data on phylogenetic estimates obtained by maximum likelihood and (Hymenoptera: Colletidae) from China with redescription of the male of C. gigas. Bayesian inference. Syst. Biol. 58, 130–145. https://doi.org/10.1093/sysbio/syp017. Zootaxa 3626, 173–187. https://doi.org/10.11646/zootaxa.3626.1.7. Leys, R., Cooper, S.J.B., Schwarz, M.P., 2002. Molecular phylogeny and historical bio- Niu, Z.Q., Zhu, C.D., Kuhlmann, M., 2013b. Bees of the Colletes clypearis-group geography of the large carpenter bees, genus Xylocopa (Hymenoptera: Apidae). Biol. (Hymenoptera: Apoidea: Colletidae) from China with description of seven new spe- J. Linn. Soc. 77, 249–266. https://doi.org/10.1046/j.1095-8312.2002.00108.x. cies. Zootaxa 3745, 101–151. https://doi.org/10.11646/zootaxa.3745.2.1. Litman, J.R., Danforth, B.N., Eardley, C.D., Praz, C.J., 2011. Why do leafcutter bees cut Niu, Z.Q., Zhu, C.D., Kuhlmann, M., 2014. Bees of the Colletes flavicornis-group from leaves? New insights into the early evolution of bees. Proc. R. Soc. B 278, 3593–3600. China with description of one new species (Hymenoptera: Apoidea: Colletidae). https://doi.org/10.1098/rspb.2011.0365. Zootaxa 3780, 534–546. https://doi.org/10.11646/zootaxa.3780.3.5. Litman, J.R., Griswold, T., Danforth, B.N., 2016. Phylogenetic systematics and a revised Noskiewicz, J., 1936. Die paläarktischen Colletes-Arten. Prace Nauk. Wydaw. Tow. Nauk. generic classification of anthidiine bees (Hymenoptera: Megachilidae). Mol. Lwowie 3, 1–531. Phylogenet. Evol. 100, 183–198. https://doi.org/10.1016/j.ympev.2016.03.018. Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess, N.D., Powell, G.V.N., Maddison, W.P., Maddison, D.R., 2018. Mesquite: A modular system for evolutionary Underwood, E.C., D’amico, J.A., Itoua, I., Strand, H.E., Morrison, J.C., Loucks, C.J., analysis. Version 3.6. http://mesquiteproject.org (accessed 8 January 2019). Allnutt, T.F., Ricketts, T.H., Kura, Y., Lamoreux, J.F., Wettengel, W.W., Hedao, P., Magnacca, K.N., Danforth, B.N., 2006. Evolution and biogeography of native Hawaiian Kassem, K.R., 2001. Terrestrial Ecoregions of the World: A New Map of Life on Earth: Hylaeus bees (Hymenoptera: Colletidae). Cladistics 22, 393–411. A new global map of terrestrial ecoregions provides an innovative tool for conserving Mardulyn, P., Whitfield, J.B., 1999. Phylogenetic signal in the COI, 16S, and 28S genesfor biodiversity. BioScience 51, 933–938. https://doi.org/10.1641/0006-3568(2001) inferring relationships among genera of Microgastrinae (Hymenoptera; Braconidae): 051[0933:TEOTWA]2.0.CO;2. evidence of a high diversification rate in this group of parasitoids. Mol. Phylogenet. Onuferko, T.M., Bogusch, P., Ferrari, R.R., Packer, L., 2019. Phylogeny and biogeography Evol. 12, 282–294. https://doi.org/10.1006/mpev.1999.0618. of the cleptoparasitic bee genus Epeolus (Hymenoptera: Apidae) and cophylogenetic Marshall, L.G., Butler, R.F., Drake, R.E., Curtis, G.H., Tedford, R.H., 1979. Calibration of analysis with its host bee genus Colletes (Hymenoptera: Colletidae). Mol. Phylogenet.

15 R.R. Ferrari, et al. Molecular Phylogenetics and Evolution 146 (2020) 106750

Evol. 141, 106603. https://doi.org/10.1016/j.ympev.2019.106603. phylogenetic inference and model choice across a large model space. Syst. Biol. 61, Packer, L., 2008. Phylogeny and classification of the Xeromelissinae (Hymenoptera: 539–542. https://doi.org/10.1093/sysbio/sys029. Apoidea, Colletidae) with special emphasis on the genus Chilicola. Syst. Entomol. 33, Roubik, D.W., 2004. Sibling species among Glossura and Glossuropoda in theAmazon re- 72–96. gion (Hymenoptera: Apidae: Euglossini). J. Kansas Entomol. Soc. 77, 235–253. Payne, A., 2014. Resolving the relationships of apid bees (Hymenoptera: Apidae) through https://doi.org/10.2317/0308.20.1. a direct optimization sensitivity analysis of molecular, morphological, and beha- Sanmartín, I., Enghoff, H., Ronquist, F., 2001. Patterns of animal dispersal, vicariance and vioural characters. Cladistics 30, 11–25. https://doi.org/10.1111/cla.12022. diversification in the Holarctic. Biol. J. Linn. Soc. 73, 345–390. https://doi.org/10. Peakall, R., Schiestl, F.P., 2004. A mark-recapture study of male Colletes cunicularius 1006/bijl.2001.0542. bees: implications for pollination by sexual deception. Behav. Ecol. Sociobiol. 56, Sann, M., Niehuis, O., Peters, R.S., Mayer, C., Kozlov, A., Podsiadlowski, L., Bank, S., 579–584. https://doi.org/10.1007/s00265-004-0816-3. Meusemann, K., Misof, B., Bleidorn, C., Ohl, M., 2018. Phylogenomic analysis of Pound, M.J., Haywood, A.M., Salzmann, U., Riding, J.B., 2012. Global vegetation dy- Apoidea sheds new light on the sister group of bees. BMC Evol. Biol. 18, 71. https:// namics and latitudinal temperature gradients during the Mid to Late Miocene (15.97- doi.org/10.1186/s12862-018-1155-8. 5.33Ma). Earth-Sci. Rev. 112, 1–22. https://doi.org/10.1016/j.earscirev.2012.02. Schmidt, S., Schmid-Egger, C., Morinière, J., Haszprunar, G., Hebert, P.D., 2015. DNA 005. barcoding largely supports 250 years of classical taxonomy: identifications for Pound, M.J., Salzmann, U., 2017. Heterogeneity in global vegetation and terrestrial cli- Central European bees (H ymenoptera, Apoidea partim). Mol. Ecol. Resour. 15, mate change during the late Eocene to early Oligocene transition. Sci. Rep. 7, 1–12. 985–1000. https://doi.org/10.1111/1755-0998.12363. https://doi.org/10.1038/srep43386. Senut, B., Pickford, M., Ségalen, L., 2009. Neogene desertification of Africa. Comptes Praz, C.J., Müller, A., Danforth, B.N., Griswold, T.L., Widmer, A., Dorn, S., 2008. Rendus Geosci. 341, 591–602. https://doi.org/10.1016/j.crte.2009.03.008. Phylogeny and biogeography of bees of the tribe Osmiini (Hymenoptera: Stamatakis, A., 2014. RAxML version 8: a tool for phylogenetic analysis and postanalysis Megachilidae). Mol. Phylogenet. Evol. 49, 185–197. https://doi.org/10.1016/j. of large phylogenies. Bioinformatics 30, 1312–1313. https://doi.org/10.1093/ ympev.2008.07.005. bioinformatics/btu033. Praz, C.J., Packer, L., 2014. Phylogenetic position of the bee genera Ancyla and Tarsalia Stephen, W.P., 1954. A revision of the bee genus Colletes in America north of Mexico (Hymenoptera: Apidae): A remarkable base compositional bias and an early (Hymenoptera, Colletidae). Univ. Kans. Sci. Bull. 36, 149–527. Paleogene geodispersal from North America to the Old World. Mol. Phylogenet. Evol. Stoeckhert, F.K., 1954. Fauna Apoideorum Germaniae. Abhandlung. Bayerisch. Akad. 81, 258–270. https://doi.org/10.1016/j.ympev.2014.09.003. Wiss. 65, 1–87. Proshchalykin, M.Y., Kuhlmann, M., 2012. The bees of the genus Colletes Latreille 1802 of Strelin, M.M., Arroyo, J.I., Fliesswasser, S., Ackermann, M., 2017. Diversification of the Ukraine, with a key to species (Hymenoptera: Apoidea: Colletidae). Zootaxa Caiophora (Loasaceae subfam. Loasoideae) during the uplift of the Central Andes. 3488, 1–40. https://doi.org/10.11646/zootaxa.3878.3.1. Org. Divers. Evol. 17, 29–41. https://doi.org/10.1007/s13127-016-0312-4. Proshchalykin, M.Y., Kuhlmann, M., 2015. Type specimens of Colletes Latreille Struwe, L., Haag, S., Heiberg, E., Grant, J.R., 2009. Andean speciation and vicariance in (Hymenoptera, Colletidae) deposited in the Zoological Institute of the Russian neotropical Macrocarpaea (Gentianaceae–Helieae). Ann. Missouri Bot. Gard. 96, Academy of Sciences, St. Petersburg, with description of a new species. Zootaxa 3949, 450–470. https://doi.org/10.3417/2008040. 540–554. https://doi.org/10.11646/zootaxa.3949.4.4. Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5: Rambaut, A., 2016. FigTree v.1.4.4. https://github.com/rambaut/figtree/releases (ac- molecular evolutionary genetics analysis using maximum likelihood, evolutionary cessed 10 May 2019). distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739. https:// Rambaut, A., Drummond, A.J., Xie, D., Baele, G., Suchard, M.A., 2018a. Posterior sum- doi.org/10.1093/molbev/msr121. marization in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, 901–904. Toro, H., 1999. Las especies Chilenas del género Colletes (Hymenoptera: Colletidae): clave https://doi.org/10.1093/sysbio/syy032. de especies. Acta Entomol. Chil. 23, 23–32. Rambaut, A., Drummond, A.J., Xie, D., Baele, G., Suchard, M.A., 2018b. Tracer v.1.7.1. Toro, H., Cabezas, V., 1977. Nuevos generos y especies de Colletini sudamericanos, se- https://github.com/beast-dev/tracer/releases/tag/v1.7.1 (accessed 10 May 2019). gunda parte. Anales Mus. Hist. Nat. Valparaiso 10, 45–64. Ramírez, S.R., Roubik, D.W., Skov, C., Pierce, N.E., 2010. Phylogeny, diversification Toro, H., Cabezas, V., 1978. Nuevos generos y especies de Colletini sudamericanos patterns and historical biogeography of euglossine orchid bees (Hymenoptera: (Apoidea Colletidae). Segunda parte. An. Mus. Hist. Nat. Valparaiso 11, 131–148. Apidae). Biol. J. Linn. Soc. 100, 552–572. https://doi.org/10.1111/j.1095-8312. Trunz, V., Packer, L., Vieu, J., Arrigo, N., Praz, C.J., 2016. Comprehensive phylogeny, 2010.01440.x. biogeography and new classification of the diverse bee tribe Megachilini: Can weuse Rasmussen, C., Cameron, S.A., 2010. Global stingless bee phylogeny supports ancient DNA barcodes in phylogenies of large genera? Mol. Phylogenet. Evol. 103, 245–259. divergence, vicariance, and long distance dispersal. Biol. J. Linn. Soc. 99, 206–232. https://doi.org/10.1016/j.ympev.2016.07.004. https://doi.org/10.1111/j.1095-8312.2009.01341.x. Upchurch, P., 2008. Gondwanan break-up: legacies of a lost world? Trends Ecol. Evol. 23, Ratnasingham, S., Hebert, P.D.N., 2007. BOLD: The Barcode of Life Data System (http:// 229–236. https://doi.org/10.1016/j.tree.2007.11.006. www.barcodinglife.org). Mol. Ecol. Notes 7, 355–364. https://doi.org/10.1111/j. Voelker, G., 1999. Dispersal, vicariance, and clocks: historical biogeography and spe- 1471-8286.2007.01678.x. ciation in a cosmopolitan passerine genus (Anthus: Motacillidae). Evolution 53, Ratnasingham, S., Hebert, P.D.N., 2013. A DNA-based registry for all animal species: the 1536–1552. https://doi.org/10.1111/j.1558-5646.1999.tb05417.x. barcode index number (BIN) system. PLoS ONE 8, 1–16. https://doi.org/10.1371/ Warncke, K., 1978. Über die westpaläarktischen Arten der Bienengattung Colletes Latr. journal.pone.0066213. (Hymenoptera, Apoidea). Pol. Pis. Entomol. 48, 329–370. Raw, A., 1984. Four new species of bees from Jamaica (Hymenoptera). Rev. Bras. Webb, S.D., 2006. The Great American Biotic Interchange: patterns and processes. Ann. Entomol. 28, 491–495. Missouri Bot. Gard. 93, 245–257. https://doi.org/10.3417/0026-6493(2006) Rech, J.A., Currie, B.S., Michalski, G., Cowan, A.M., 2006. Neogene climate change and 93[245:TGABIP]2.0.CO;2. uplift in the Atacama Desert, Chile. Geology 34, 761–764. https://doi.org/10.1130/ Weir, J.T., Price, M., 2011. Andean uplift promotes lowland speciation through vicariance G22444.1. and dispersal in Dendrocincla woodcreepers. Mol. Ecol. 20, 4550–4563. https://doi. Rech, J.A., Currie, B.S., Shullenberger, E.D., Dunagan, S.P., Jordan, T.E., Blanco, N., org/10.1111/j.1365-294X.2011.05294.x. Tomlinson, A.J., Rowe, H.D., Houston, J., 2010. Evidence for the development of the Wiens, J.J., 2003. Missing data, incomplete taxa, and phylogenetic accuracy. Syst. Biol. Andean rain shadow from a Neogene isotopic record in the Atacama Desert, Chile. 52, 528–538. https://doi.org/10.1080/10635150390218330. Earth Planet. Sci. Lett. 292, 371–382. https://doi.org/10.1016/j.epsl.2010.02.004. Woodburne, M.O., 2010. The Great American Biotic Interchange: Dispersals, Tectonics, Ree, R.H., Moore, B.R., Webb, C.O., Donoghue, M.J., 2005. A likelihood framework for Climate, Sea Level and Holding Pens. J. Mamm. Evol. 17, 245–264. https://doi.org/ inferring the evolution of geographic range on phylogenetic trees. Evolution 59, 10.1007/s10914-010-9144-8. 2299–2311. https://doi.org/10.1111/j.0014-3820.2005.tb00940.x. Yu, Y., Harris, A.J., Blair, C., He, X., 2015. RASP (Reconstruct Ancestral State in Ree, R.H., Sanmartín, I., 2018. Conceptual and statistical problems with the DEC+J Phylogenies): a tool for historical biogeography. Mol. Phylogenet. Evol. 87, 46–49. model of founder-event speciation and its comparison with DEC via model selection. https://doi.org/10.1016/j.ympev.2015.03.008. J. Biogeogr. 45, 741–749. https://doi.org/10.1111/jbi.13173. Zachos, J.C., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, Rhythms, and Ree, R.H., Smith, S.A., 2008. Maximum-likelihood inference of geographic range evolu- Aberrations in Global Climate 65 Ma to Present. Science 292, 686–694. https://doi. tion by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4–14. https://doi. org/10.1126/science.1059412. org/10.1080/10635150701883881. Zachos, J.C., Dickens, G.R., Zeebe, R.E., 2008. An early Cenozoic perspective on green- Ronquist, F., Teslenko, M., Van Der Mark, P., Ayres, D.L., Darling, A., Höhna, S., Larget, house warming and carbon-cycle dynamics. Nature 451, 279–283. https://doi.org/ B., Liu, L., Suchard, M.A., Huelsenbeck, J.P., 2012. MrBayes 3.2: efficient Bayesian 10.1038/nature06588.

16